Our data were derived from the RNA-seq data obtained from the COAD project of the TCGA database (https://portal.gdc.cancer.gov/). Gene expression, baseline data, clinical logistic correlation, prognostic analysis, clinical diagnostic value, and gene function clustering data were obtained through the R package. Our study excluded samples with insufficient “0” gene expression values and survival information. We retained the RNA-seq and clinical data for further study. A total of 478 COAD patients with comparable clinical symptoms were included in this investigation. Our study conforms with the publishing criteria established by TCGA.

Tumor Immune Estimation Resource (TIMER) (https://cistrome.shinyapps.io/timer/) was utilized to investigate the possibility of a link between MAD2L2 expression and tumor-infiltrating immune cells. We used a genetic module to determine the relationship of MAD2L2 in COAD with tumor-infiltrating immune cells, including CD4+ T cells, dendritic cells, B cells, neutrophils, and macrophages. TIMER developed a graph demonstrating the association between the degree of gene expression and the purity of the tumor. Furthermore, we investigated the connection between MAD2L2 and 24 immune-infiltrating cells in COAD. The markers for 24 immune cells were taken from an article in Immunity (Bindea et al., 2013).

The colon cancer cell line HCT116, used in the experiments, was obtained from the American Type Culture Center. Cells were cultured in DMEM (Vivacell; Shanghai, China) supplemented with 10% FBS (Vivacell; Shanghai, China) and penicillin-streptomycin (10,000 U/ml; Beyotime Institute of Biotechnology). Culture was performed at 37°C and 5% CO₂. When HCT116 cells reached 70% growth status, proteins were collected after treatment with TNF-α cytokines at different concentrations (0, 20, 20, 40, 60, 80, 100 ng/ml) for 24 h. TNF-α 50 mg (Peprotech, Suzhou, China).

Western blotting was used to isolate and identify proteins. Total proteins were extracted using RIPA buffer (Beyotime Institute of Biotechnology, Shanghai, China). Protein concentration was then determined using the BCA kit (Nanjing Capgemini Biotechnology Co., Ltd., China). Proteins (10–20 µg/well) were passed through 10% SDS-PAGE at 80V, and then transferred onto 0.2 µm PVDF membranes under wet conditions at 200 mA. The blotted membranes were blocked with 5% skim milk for 1 h at room temperature. The antibody was diluted in primary antibody diluent. Primary antibodies were incubated at room temperature for 3 h. Secondary antibodies were incubated for 1 h at room temperature. Chemiluminescence signal was detected using Pierce™ ECL and analyzed by Image Lab software (version 5.2.1 Bio-Rad Laboratories, Inc., BIO-RAD, CA, USA). The following antibodies were used: primary antibody-TNF-α (cat. 8184s; 1:1,000; cell signaling, USA), MAD2L2 (cat. ab180579; 1:1,000; Abcam, UK), GAPDH (cat. AB-P-R001; 1:1,000; Hangzhou Xianzhi Biotechnology Ltd., China); and secondary antibody-goat anti-rabbit IgG H&L (HRP) (cat. ZB-2306; 1:10,000; Beijing Zhongshan Jinqiao, China).

Transwell assay was used to detect cell invasion and migration. Cell migration was treated by taking cells in logarithmic growth phase and digesting the cells with trypsin. A 24-well plate was taken, 600 ul of complete medium was added to the lower chamber, and the transwell was placed in the plate and incubated for 48 h. Afterwards, 4% paraformaldehyde was added to fix the cells for 20 min, and then the cells were washed 3 times with PBS, stained with crystal violet for 15 min, and photographed and cell counted under a microscope (200X). The cell invasion was treated by first laying down the matrix gel. After that, a 24-well plate was taken, 600 ul of medium containing chemokine was added to the lower chamber, the transwell chambers after gelling were put into the well plate, 100–200 ul of cell suspension was taken and mixed and added to the upper chamber, and the culture was fixed with 4% paraformaldehyde after 48 h, stained with crystalline violet for 15 min, and Matrigel and the non-migrated cells in the upper chamber were gently wiped with cotton swabs, PBS-washed, photographed and cell counted under a microscope (200X). The following materials were mainly used, transwell chambers (Corning Inc., USA), and Matrigel gel (BD Inc., USA).

All statistical analyses were performed using R package (version 3.6.3). To calculate MAD2L2 expression, the Mann–Whitney U test was used. The correlation between clinical features and MAD2L2 expression was analyzed using a binary logistic model. In calculating the 95% CI and HR, the Cox regression module to analyze univariate and multivariate models was used. To compare numerous clinical parameters and survival rates, a univariate survival analysis was performed on each individual patient. Using multivariate Cox analysis, we were able to determine the expression of MAD2L2 as well as the presence of additional pathological and clinical variables. The threshold for MAD2L2 expression was established at a p-value of less than 0.05.

In February 2022, after screening, a total of 478 patients with clinical features related to MAD2L2 expression were obtained from the TCGA website. Table 1 lists the detailed clinical characteristics. Among the 478 participants, among those with low MAD2L2 expression, 121 were female (25.3%), and 118 were male (24.7%); among those with high expression of MAD2L2, 105 were female (22%), and 134 were male (28%). The age of all participants was cut off at 65 years. In terms of COAD pathological stage, among patients with low MAD2L2 expression, 39 patients were in Stage I (8.4%), 89 patients were in Stage II (19.1%), 69 patients were in Stage III (14.8%), and 37 patients were in Stage IV (7.9%); among patients with high MAD2L2 expression, 42 patients were in Stage I (9%), 98 patients were in Stage II (21%), 64 patients were in Stage III (13.7%), and 29 patients were in Stage IV (6.2%). History of colonic polyps, low expression of MAD2L2, 149 patients had no history of colonic polyps (36.5%) and 53 patients had a history of polyps (13%); among the highly expressed MAD2L2, 113 patients had no history of colon polyps (27.7%) and 93 patients had a history of colon polyps (22.8%). The median follow-up time among the 478 patients was 27.6 months.

MAD2L2 expression was increased in colon cancer, adrenocortical carcinoma, uroepithelial carcinoma of the bladder, invasive breast cancer, squamous and adenocarcinoma of the cervix, bile duct cancer, diffuse large B-cell lymphoma, esophageal cancer, glioblastoma multiforme, squamous cell carcinoma of the head and neck, low-grade (Fig. 1). RNA-seq research indicated that MAD2L2 expression was much greater in COAD cancer tissues than in paraneoplastic tissues in unpaired vs. paired samples (Figs. 2A and 2B). MAD2L2 mRNA expression was considerably elevated in COAD, as revealed by the findings.

Table 2 summarizes the correlation between MAD2L2 expression and clinical logistic characteristics in patients with COAD. Expression of MAD2L2 was correlated with the history of colon polyps (p < 0.001) but not with other clinical features. Meanwhile, in the box plot of clinical features, we found that MAD2L2 expression correlated with tumor status and history of colon polyps (Figs. 3A–3H). The results of a univariate study using Cox regression revealed that MAD2L2 gene expression was a categorical dependent variable that was related with poor prognostic clinical characteristics (Table 3). The expression of MAD2L2 was significantly correlated with age greater than 65 years (p = 0.028, [CI] = 1.052–2.463). N stage (N1: p = 0.042, [CI] = 1.019–2.771, N2: p < 0.001, [CI] = 2.593–6.329), M stage (M1: p < 0.001, [CI] = 2.683–6.554), pathological stage (Stage III: p = 0.007, [CI] = 1.436–9.448, Stage IV: p < 0.001, [CI] = 3.608–23.936), CEA level (greater than 5: p < 0.001, [CI] = 1.788–5.471) are relevant. These results suggest that high MAD2L2 expression is highly correlated with poor clinical prognosis in COAD. We also examined the effect of high MAD2L2 expression on the migration and invasion of COAD cells HCT116 and found that the migration and invasion expression of HCT116 cells were significantly increased when MAD2L2 was highly expressed (Figs. 3I and 3J). This result indicates further validation that, high MAD2L2 expression causes poor prognosis of COAD.

The gene expression data for MAD2L2 was used to conduct a ROC curve analysis to determine the clinical diagnostic value of this gene. As shown in Fig. 4A, the AUC area of MAD2L2 was 0.93 when compared to paracancerous tissue vs. COAD. Using subgroup analysis, it was discovered that MAD2L2 gene expression was diagnostic in different stages of COAD, with AUC values of 0.922 in paracancerous tissue vs. stage I, 0.744 in T1/T2, 0.699 in T1/T3, 0.744 in T1/T4, and 0.744 in paracancerous tissue vs. N0 stage, paracancerous tissue vs. M0 stage, and paracancerous tissue vs. M1 stage being the most significant (Figs. 4B–4H). The results showed that MAD2L2 has certain diagnostic value in different clinical stages. We developed a nomogram to predict the 1-, 3-, and 5-year survival probability of patients with COAD by integrating the MAD2L2 gene expression level with characteristics related with clinical prognosis and diagnosis (Fig. 5). The score could be used by doctors to forecast patients’ 1-, 3-, and 5-year survival rates. These findings indicate that MAD2L2 could be utilized as a predictor in the diagnosis of COAD at various clinical stages.

Tumor-infiltrating immune cells play a key role in predicting overall survival in patients with COAD. Therefore, using TIMER, we analyzed the correlation between MAD2L2 expression and immune infiltration levels in COAD. As shown in Fig. 6A, MAD2L2 expression was positively correlated with neutrophil levels (p = 7.4 × 10⁻⁴). The results suggest that MAD2L2 plays a role in the immune infiltration of COAD. Additionally, we aimed to evaluate whether the tumor immune microenvironment of patients with COAD who expressed high levels of MAD2L2 was distinct from that of patients with COAD who expressed low levels of MAD2L2. The 480 COAD samples were divided into two groups based on MAD2L2 expression, with 240 samples classified as high expression and 240 as low expression. Using the R package, we performed calculations on immune infiltration to determine levels of 24 immune cells. The immune infiltration algorithm applied to 24 immune cell subtypes helped to assess differences in immune cell expression levels between the high and low MAD2L2 expression groups (Fig. 6B). T cells, aDCs, cytotoxic cells, DCs, iDCs, neutrophils, NK CD56 bright cells, NK CD56dim cells, Tcm cells, Th1 cells, Th17 cells, and Tregs were significantly affected by MAD2L2 expression. Compared with the low expression group, T cells, aDCs, cytotoxic cells, DCs, iDCs, neutrophils, NK CD56 bright cells, NK CD56dim cells, Th1 cells, Th17 cells, and TReg increased in the high expression group (p < 0.05), while Tcm decreased (p < 0.05). We also assessed possible correlations between 24 immune cells (Fig. 6C). The generated heatmap revealed associations between the ratios of several tumor-infiltrating immune cell groups ranging from weak to robust. Also, we validated it at the protein level. Using different concentrations of TNF-α acting on colon cancer HCT116 cells, MAD2L2 showed a dose-dependent effect as the concentration of TNF-α increased (Figs. 7A and 7B). As a result, we hypothesize that MAD2L2 expression is associated with tumor immune infiltration throughout the development of COAD and, to a lesser degree, influences immune cell expression in the tumor immune microenvironment.

To understand the biological significance of MAD2L2 in COAD, we used the R package to analyze the co-expression network of MAD2L2 in COAD data. Based on our findings, we screened and examined the influence of the expression of the top 50 genes positively related with MAD2L2, against the top 50 genes negatively associated with MAD2L2 expression, on overall survival in colon cancer. The results are shown in the heat map in Figs. 8A and 8B. Genetic risk for COAD is predicted to be highest in the top 50 positively associated genes. The top 50 negatively related genes, on the other hand, were the ones that were most likely to be low-risk genes for COAD, according to the analysis. Additionally, this shows that MAD2L2, along with other genes, may be implicated in the overexpression of COAD risk factors and the downregulation of COAD preventative factors, as well as in the occurrence and development of COAD. The Kyoto Encyclopedia of Genes and Genomes (KEGG) Gene Ontology (GO) term annotations showed that MAD2L2 was mainly involved in biological processes when enriched with positively correlated genes, including response to hypoxia, response to decreased oxygen levels, anaphase-promoting complex-dependent catabolic process, response to oxygen levels, mitochondrial translational elongation, cellular components including vesicle coat, Fanconi anemia nuclear complex, ribosome, mitochondrial matrix, mitochondrial inner membrane, pathways including Huntington’s disease and Parkinson’s disease, ubiquitin mediated proteolysis, pathways of neurodegeneration-multiple diseases (Fig. 8C). Histone modification and covalent chromatin modification were inhibited, as were molecular functions such as histone binding, modification-dependent protein binding, histone demethylase activity (H3-K9 specific), methylation-dependent protein binding, and lysine degradation. Biological processes such as histone modification and covalent chromatin modification were also inhibited, as well as pathways such as lysine degradation, which were all inhibited as well (Fig. 8D). From these results, we infer that in COAD, processes such as the hypoxia response, response to reduced oxygen levels, and mitochondrial translation elongation are closely related to MAD2L2 expression.