TCGA tumor dataset for 33 cancer types was downloaded from https://genome-cancer.ucsc.edu/. For differential expression analysis, the R (version “4.1”) package ‘limma’ was used (12). Oncomine, a cancer microarray database and web-based data-mining platform aimed at facilitating discovery from genome-wide expression analyses (https://www.oncomine.org/resource/login.html). Search for “TTC35”, threshold (p-value): 0.05, threshold (fold change): 1.5, threshold (gene rank): Top 10%, data type: all (13). A p-value < 0.05 was considered statistically significant.

Gene expression profiling interactive analysis 2 (GEPIA2; http://gepia2.cancer-pku.cn/#index/) is a tumor/normal differential expression analysis based on TCGA and gene tissue expression (GTEx) data, for analysis according to cancer type or pathological stage, patient survival analysis, network tools for similar gene detection, correlation analysis, and dimensionality reduction analysis (14). GEPIA2 was used to detect the expression of EMC2 and lncRNA in different types of human cancers, with p < 0.05 considered statistically significant. It was also used to analyze the survival of EMC2 in 33 different cancer types (including overall survival [OS] and relapse-free survival [RFS]), as well as to evaluate the prognostic value of candidate lncRNAs in BRCA. A log-rank p-value < 0.05 was considered statistically significant. In addition, the GEPIA2 database was used to evaluate the correlation between EMC2 expression in BRCA and immune checkpoints. Absolute values of R > 0.1 and p < 0.05 were judged as statistically significant. Univariate and multivariate Cox proportional hazards models were used for survival analysis, and the R package ‘Survival’ (2.41-3) was used for the Kaplan–Meier estimation of OS.

After logging in to the cBioPortal website (https://www.cbioportal.org/), “TCGA Pan-Cancer Atlas Research” was selected in the “Quick Selection” column, and “EMC2” was entered to query gene mutation characteristics (15). We observed the mutation frequency, mutation type, and copy number variation (CNV) results of all TCGA tumors in the “Tumor Type Summary” module, and observed the mutation site information of EMC2 in the protein structure diagram through the “mutation” module. We also used the “comparison” module to obtain data on overall, disease-free, progression-free, and disease-free survival differences in TCGA cancer cases with or without EMC2 genetic changes. Kaplan–Meier plots of log-rank p-values were also generated.

Shiny methylation analysis resource tool (SMART; http://www.bioinfo-zs.com/smartapp/) is a user-friendly and easy-to-use web application for comprehensively analyzing the DNA methylation data of TCGA (16). Here, by searching “EMC2,” we explored the difference in the average methylation level of the tumor and normal groups of EMC2 in each type of cancer and the effect of methylation on tumor survival.

Several target gene prediction programs, such as PitA, RNA22, miRmap, microT, Miranda, PicTar, and TargetScan, were used to predict the upstream binding miRNAs of EMC2 (17). Only predicted miRNAs usually appear in the above two programs before they are used in subsequent analyses. These predicted miRNAs were considered candidate miRNAs for EMC2.

starBase (http://starbase.sysu.edu.cn/) is the most comprehensive database of mRNA and protein-RNA interaction networks (18). We used starBase to analyze the expression correlation of miRNA-EMC2, hsa-miR-410-3p-lncRNA, or lncRNA-EMC2 in BRCA. TCGA data was used to analyze the expression levels of hsa-miR-410-3p and lncRNA in BRCA and normal controls.

The online Kaplan–Meier plotter (http://kmplot.com/analysis/) was used to perform a hsa-miR-410-3p survival analysis on BRCA, and a log-rank p-value < 0.05 was considered statistically significant (19).

The tumor immune estimation resource (TIMER, https://cistrome.shinyapps.io/timer/) is a web server for the comprehensive analysis of tumor-infiltrating immune cells (20). TIMER is used to analyze the correlation between the expression level of EMC2 in BRCA and the expression level of immune checkpoints, and used the R package ‘GSVA’ to analyze the relationship between the expression level of EMC2 and immune cell infiltration in breast cancer. p < 0.05 was considered statistically significant.

The human breast epithelial cell line MCF10A and human breast cancer cell lines MCF-7 and MDA-MB-231 were obtained from Bei Na Biotechnology Co., Ltd. (Beijing, China). The MCF7 and MDA-MB-231 cells were maintained in high-glucose Dulbecco’s modified Eagle medium (DMEM; Gibco, Grand Island, NY, USA) supplemented with 10% fetal bovine serum (FBS; Gibco), and 1% penicillin–streptomycin (Gibco). The MFC10A cells were cultured in DMEM/Ham’s F-12 medium (Invitrogen) supplemented with 10% FBS (Gibco), 0.5 µg/ml hydrocortisone (Sigma-Aldrich, St. Louis, MO, USA), 20 ng/ml epidermal growth factor (Sigma-Aldrich), 10 μg/ml insulin (Sigma-Aldrich), and 1% penicillin–streptomycin (Gibco). Cells were cultured in a humidified atmosphere containing 5% CO₂ and 95% air at 37°C.

Total RNA was extracted from the normal human breast epithelial cell line MCF 10A and breast cancer cell lines MCF7 and MDA-MB-231 using TRIzol reagent (Sigma-Aldrich). The extracted RNA was reverse-transcribed to cDNA using HiScript II Q RT SuperMix (R223-01; Vazyme Biotech, Nanjing, China). Real-time PCR was performed on a QuantStudio™ 6 Flex (Life Technologies, USA) using ChamQ SYBR qPCR Master Mix (Q711-02; Vazyme Biotech, Nanjing, China). The 2^–ΔΔCt method was used to calculate the relative gene expression levels and then normalized against GAPDH. Primers were designed and synthesized by Sangon Biotech (Shanghai, China). The primer sequences are listed in Supplementary Table 1 .

The statistical analysis in this study was automatically calculated by the online databases mentioned in the above sections. A p-value < 0.05 or log-rank p-value < 0.05 was considered to be statistically significant.

To explore the possible role of EMC2 in carcinogenesis, we first analyzed its expression in 33 human cancers. As shown in Figure 1A , compared with the normal group, EMC2 was expressed differently in 22 cancer types, of which 18 were highly expressed in the tumor group, that included breast invasive carcinoma (BRCA), cholangiocarcinoma (CHOL), colon adenocarcinoma (COAD), lymphoid neoplasm diffuse large B-cell lymphoma (DLBC), glioblastoma multiforme (GBM), head and neck squamous cell carcinoma (HNSC), kidney chromophobe (KICH), kidney renal clear cell carcinoma (KIRC), brain lower grade glioma (LGG), liver hepatocellular carcinoma (LIHC), lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), pancreatic adenocarcinoma (PAAD), prostate adenocarcinoma (PRAD), rectum adenocarcinoma (READ), stomach adenocarcinoma (STAD), skin cutaneous melanoma (SKCM), and thymoma (THYM). There were four cancer types with low expression in the tumor group, comprising acute myeloid leukemia (LAML), ovarian serous cystadenocarcinoma (OV), testicular germ cell tumors (TGCT), and thyroid carcinoma (THCA). However, BRCA, CHOL, DLBC, GBM, HNSC, KIRC, LAML, LIHC, LUAD, LUSC, OV, PAAD, PRAD, SKCM, STAD, TGCT, and THYM showed significant differences in 17 tumors. We further verified the expression of EMC2 in tumors containing paired samples. As shown in Figure 1B , there were significant differences in the expression levels of EMC2 in BRCA, CHOL, esophageal carcinoma (ESCA), HNSC, KIRC, papillary renal cell carcinoma (KIRP), LIHC, LUSC, STAD, and THCA compared with the corresponding adjacent normal tissue controls. Second, we used GEPIA2 to analyze the influence of EMC2 expression on tumor grading and found significant differences in COAD, ESCA, KIRC, and TCGT ( Figure 1C ). The Oncomine database was used to evaluate the differential expression of EMC2 in tumors, and similar results were obtained ( Supplementary Figure 1 ). The above results show that EMC2 is a potential oncogene for eight cancers.

Next, a pan-cancer survival analysis was performed; it included two prognostic indices: OS and RFS. For OS, a high EMC2 expression level showed unfavorable prognoses in bladder urothelial carcinoma (BLCA), BRCA, and uveal melanoma (UVM); however, KIRC patients with higher EMC2 expression showed a better prognosis ( Figure 2A ). For RFS, BLCA and HNSC patients with high EMC2 expression levels showed poor prognoses, while KIRC and pheochromocytoma and paraganglioma (PCPG) patients with high EMC2 expression levels had better prognostic survival ( Figure 2B ). On this basis, we used univariate and multivariate Cox regression to analyze the effect of EMC2 expression and other clinicopathological factors on breast cancer prognosis. The top 300 positive and negative correlation genes of EMC2 were then subjected to GO and KEGG enrichment analyses, respectively. As shown in Supplementary Table 2 , the results show that EMC2 expression had an important effect on breast cancer and could be used as an independent prognostic indicator; these findings were irrespective of the regression analysis used (univariate or multivariate). Therefore, the expression of EMC2 in breast cancer can have significant effects on factors such as tumor progression and even patient survival. GO and KEGG enrichment analysis found that genes positively related to EMC2 are mainly involved in RNA transport and cell cycle regulation ( Supplementary Figure 2 ), whereas genes negatively related to EMC2 are mainly involved in ECM-receptor interaction, the relaxin signaling pathway, and focal adhesion ( Supplementary Figure 3 ).

We observed EMC2 mutations in different tumor samples in TCGA cohort. As shown in Figure 3A , the frequency of EMC2 mutations in OV patients with “amplification” as the primary type was the highest (>15.41%), followed by breast cancer (10.33%). Uterine corpus endometrial carcinoma (UCEC) was mainly of the “mutation” type, and its mutation frequency was about 4.54%. Some patients (about 1.15%) with mesothelioma lacked EMC2. Figure 3B shows the types and locations of EMC2 mutations. In addition, we explored the potential relationship between EMC2 mutations and the clinical survival and prognosis among patients with different cancer types. Figure 3C shows that compared with PRAD patients without EMC2 mutations, those with EMC2 mutations had decreased OS (p = 3.384e-4), disease-free survival (p = 1.137e-3), and progression-free survival (p = 8.209e-4), whereas the disease-specific survival rate (p = 2.81e-7) was particularly poor. In breast cancer patients, survival was not affected by the above mentioned mutations. Considering the number of clinical samples, the effect of EMC2 structural mutations on breast cancer prognosis requires further analysis.

DNA methylation can function to silence tumor suppressor genes along with genetic mutations. Thus, we calculated differentially methylated CpG sites based on the tumor and normal samples. The CpG sites of the EMC2 promoter were significantly methylated in the tumor samples compared to in the normal samples of most cancer types ( Figure 4A ). Moreover, EMC2 methylation had a significant effect on the survival of tumors ( Figure 4B ). We found that EMC2 showed hypomethylation in BLCA, BRCA, HNSC, KIRC, KIRP, LIHC, LUAD, PRAD, READ, and UCEC, but hypermethylation in COAD and LUSC.

To determine whether ncRNAs regulate EMC2, we first predicted upstream miRNAs that could bind to EMC2, and found 19 such miRNAs. To improve visualization, a miRNA–EMC2 regulatory network was established using Cytoscape software ( Figure 5A ). Based on the mechanism of miRNA in regulating the target gene expression, we speculated a negative correlation between miRNA and EMC2; therefore, an expression correlation analysis was performed. As shown in Figure 5B , EMC2 expression was significantly negatively correlated with miR-410-3p expression. The expression and prognostic value of miR-410-3p in BRCA were also determined. As mentioned above, in Figures 5C, D , miR-410-3p expression was significantly downregulated in BRCA, and its upregulated expression was positively correlated with prognosis ( Figure 5E ). These findings could that miR-410-3p is the most significant miRNA regulating EMC2 in BRCA.

Using starBase, 30 lncRNAs were predicted that were upstream of miR-410-3p. Similarly, to improve visualization, the lncRNA-miR-410-3p regulatory network was constructed using Cytoscape software ( Supplementary Figure 4 ). Then, the corresponding normal tissue data in BRCA were extracted using GTEx and TCGA to determine the expression levels of these lncRNAs in BRCA. As shown in Figures 6A–D , among all the 30 lncRNAs, only NEAT1 and LINC00665 were significantly down- and upregulated in BRCA, respectively, compared to the normal control. Subsequently, the prognostic value of the two lncRNAs in BRCA was evaluated. As shown in Figures 6E–H , these lncRNAs did not significantly influence the survival of breast cancer patients. According to the competing endogenous RNA (ceRNA) hypothesis, lncRNA can increase mRNA expression by competitively binding and sharing miRNA. Therefore, a negative correlation between lncRNA and miRNA or a positive correlation between lncRNA and mRNA plausibly exists. As shown in Table 1 , the starBase database was used to detect the expression correlation between the two lncRNAs in BRCA and miR-410-3p or EMC2. Using the expression analysis, survival analysis, and considering the correlation analysis, LINC0065 was found the most significant lncRNA upstream of the miR-410-3p/EMC2 axis in BRCA.

Immunotherapy is a treatment method that enhances or inhibits immune system function to cure diseases. Immunotherapy has achieved satisfactory results in malignant melanoma, lung cancer, and other tumors. Compared with lung cancer and melanoma, breast cancer is considered a “weakly immunogenic” tumor, and its immunotherapy effect is not sufficient. Therefore, it is necessary to find new genes to improve the immunotherapy of breast cancer. As correlation analysis can provide critical clues for studying EMC2 functions and mechanisms, the correlation between the expression level of EMC2 and the level of immune cell infiltration was evaluated. As shown in Figure 7A , the expression of EMC2 in BRCA was related to the infiltration of 17 immune cells, including CD8⁺ T cells, mast cells, natural killer (NK) cells, T helper cells (Th1, Th2, and Th17), and neutrophils.

To further explore the role of EMC2 in tumor immunity, we used the GEPIA database to determine the correlation of EMC2 expression with immune cell biomarkers in BRCA. As shown in Table 2 , EMC2 showed correlation with B-cell biomarkers (CD19 and CD79A), CD8⁺ T cell biomarkers (CD8B), M1 macrophage biomarkers (IRF5), M2 macrophage biomarkers (CD163 and MS4A4A), dendritic cell biomarkers (HLA-DPB1, CD1C, and NRP1), NK cell biomarkers (KIR2DL3, KIR2DL4, and KIR3DL3), mast cell biomarkers (TPSB2 and TPSAB1), Th1 cell markers (TNF-α), Th2 cell markers (GATA3), Th17 cell markers (STAT3), and Tfh cell markers (BCL6). These findings partially indicate a significant correlation between EMC2 expression and immune cell infiltration.

PD1/PD-L1, CTLA-4, and TIM-3 are critical immune checkpoints responsible for tumor immune escape. Considering the potential carcinogenic effects of EMC2 in BRCA, the relationships between BRCA and PD1, PD-L1, CTLA-4, and TIM-3 were evaluated. Figures 7B–E shows that EMC2 expression is significantly positively correlated with PD-L1, CTLA-4, and TIM-3 in BRCA, regulated by purity. From the expression correlation analysis, we also found that EMC2 was significantly positively correlated with PD-L1, CTLA-4, and TIM-3 in BRCA ( Figures 7F–I ). These results indicate that tumor immune escape may involve the EMC2-related carcinogenesis of BRCA.

After analyzing EMC2 expression with RT-qPCR ( Figure 8A ) and immunohistochemistry ( Figures 8B–D ), we confirmed that it was upregulated in MCF7 and MDA-MB-231 cells compared to MCF10A cells. The HPA provides a map showing the distribution and relative abundance of proteins in normal human breast tissues and cancer tissues. The protein levels of the EMC2 was significantly higher in tumor tissues compared with normal tissues based on the HPA database; the EMC2 protein levels were consistent with the RT-qPCR results. The results were significant (p < 0.05).