3.1 Finding genes linked to mitochondria that are differently expressed in epilepsy
The MitoCarta 3.0 database granted us access to the mitochondria-related gene set, which encompasses 1136 human mitochondrial genes. The expression of these genes was extracted from the GSE28674 gene set, and a matrix of gene expression associated with mitochondria was constructed and analyzed for differential expression profiling. The analysis revealed the identification of 82 MitDEGs (Supplementary Data Sheet 2), comprising 30 up-regulated genes and 52 down-regulated genes (Fig. 1A). Subsequently, we delved into the potential signaling pathways associated with these MitDEGs through GO and KEGG enrichment analyses (Supplementary Data Sheet 3). BP enrichment analyses (Fig. 1B) showed that MitDEGs were involved in the small molecule metabolic process, oxoacid metabolic process, and mitochondrial gene expression. CC enrichment analyses (Fig. 1C) showed that MitDEGs mainly functioned in mitochondria. BP enrichment analysis (Fig. 1B) showed that all MitDEGs were involved in small molecule metabolic process, oxoacid metabolic process, and mitochondrial gene expression, and CC enrichment analysis (Fig. 1C) indicated that MitDEGs mainly functioned in mitochondria, and MF enrichment analysis (Fig. 1D) showed that MitDEGs have mainly participated in catalytic activity, coenzyme binding and mitochondrial gene expression. KEGG analysis (Fig. 1E) showed that MitDEGs were mainly engaged in Metabolic pathways and the Citrate cycle (TCA cycle).
3.2 Identification of the Hub gene
We employed a machine learning approach to screen for hub genes, aiming to identify key hub genes associated with mitochondrial-related genes in epilepsy. The list of 82 MitDEGs that were initially linked to epilepsy was reduced to 8 MitDEGs that are most significantly connected with epilepsy using the LASSO regression approach (Fig. 1F). A diagnostic model for classifying epilepsy constructed on the 8 hub genes was created using a logistic regression approach. The outcomes demonstrated that epilepsy patients may be distinguished from normal samples using the classification diagnostic model built using eight Hub genes. The hub genes' and models' AUC values (Fig. 1G, H): AUC = 0.806 for PHB, AUC = 0.931 for ACADSB, AUC = 0.944 for CCDC58, AUC = 0.903 for CKMT1A, AUC = 0.903 for PISD, AUC = 0.944 for STAR, AUC = 0.861 for TOMM20, AUC = 0.861 for PDE2A, and AUC = 1.000 for the model. The results showed that the CCDC58 gene as a diagnostic marker is able to diagnose epilepsy patients well, and CCDC58 has been rarely reported in epilepsy as well as other diseases. Therefore, we selected CCDC58 as a key molecule for subsequent pan-cancer analysis.
3.3 Expression of CCDC58 and its cellular location in various malignancies and healthy tissues
Our findings demonstrated elevated CCDC58 mRNA expression levels in 26 out of 33 malignancies compared to normal tissues (Fig. 2A). Additionally, we assessed CCDC58 expression levels in adjacent normal tissues and matched cancer tissues in patient malignancies using the TCGA dataset. We found that 13 out of 23 cancers had considerably greater expression of CCDC58 compared to normal tissues (Fig. 2B). As further analyzed by Human Protein Atlas (HPA) database, CCDC58 protein was found in Lung cancer (11/11), Colorectal cancer (10/10), Head and neck cancer (4/4), Stomach cancer (11/11), Carcinoid (4/4), Pancreatic cancer (10/10), Prostate cancer (11/11), Breast cancer (12/12) and Melanoma (11/11) tumor types with 100% expression frequency, while in Thyroid cancer (3 /4), Lymphoma (9/12) and Glioma (8/12) tumor types with the lowest expression frequency (Fig. 2C). Histopathology showed that CCDC58 was mainly localized to mitochondria within cells (Fig. 2D).
3.4 Pan-cancer analysis of the prognostic value of CCDC58
We conducted survival studies for 36 malignancies using the following metrics: overall survival (OS), disease-free survival (DSS), disease-free interval (DFI), and progression-free interval (PFI) in order to further study the relationship between CCDC58 expression and prognosis. CCDC58 expression was found to be substantially (p < 0.05) linked with overall survival in ACC, KICH, LIHC, PAAD, PRAD, and SARC, according to univariate Cox analysis (Fig. 3A). Additionally, Kaplan-Meier analysis revealed a negative correlation between survival time and CCDC58 expression in ACC (Fig. 3B), KICH (Fig. 3C), LIHC (Fig. 3D), PAAD (Fig. 3E), PRAD (Fig. 3F), and SARC (Fig. 3G).
Moreover, the DSS analysis revealed a significant correlation (p < 0.05) between the expression of CCDC58 and the prognosis of KICH, LIHC, PAAD, and PRAD (Figure S1A). High CCDC58 expression was inversely correlated with a poor prognosis for KICH (Figure S1B), LIHC (Figure S1C), PAAD (Figure S1D), and PRAD (Figure S1E), according to Kaplan-Meier analysis. We observed a significant association between high CCDC58 expression and reduced Disease-Free Interval (DFI) in LIHC, KIRP, ACC, KIPAN, COAD, COADREAD, and TGCT (Figure S2A). Furthermore, Kaplan-Meier analysis revealed that high CCDC58 expression was linked to decreased DFI in ACC (Figure S2B), COAD (Figure S2C), COADREAD (Figure S2D), KIPAN (Figure S2E), KIRP (Figure S2F), LIHC (Figure S2G), and TGCT (Figure S2H).
Forest plots revealed a significant association between CCDC58 expression and PFI in ACC, KICH, KIPAN, KIRP, LIHC, PAAD, and PRAD (Figure S3A). KM survival analysis also demonstrated a significant association between CCDC58 expression and reduced PFI in ACC (Figure S3B), KICH (Figure S3C), KIPAN (Figure S3D), KIRP (Figure S3E), LIHC (Figure S3F), PAAD (Figure S3G), and PRAD (Figure S3H) patients. Upon concluding the analysis of these four prognostic datasets, we observed that CCDC58 exhibited significant prognostic value in urological tumors, particularly kidney renal clear cell carcinoma and kidney chromophobe. Our findings suggest a strong correlation between CCDC58 expression and the prognosis of patients with urological tumors.
3.5 Prognostic value of CCDC58 expression in gliomas
We conducted the following survival studies to look into the relationship between CCDC58 expression and glioma prognosis: overall survival (OS), disease-free survival (DSS), and disease-free interval (DFI). Elevated CCDC58 expression was associated with poorer overall survival (Fig. 4A), disease-specific survival (Fig. 4B), and disease-free interval (Fig. 4C), as indicated by Kaplan-Meier analyses. Furthermore, we looked into the relationship between prognosis (OS) and CCDC58 expression in several clinical subgroups of gliomas within the TCGA database. In the subsequent clinical subgroups, the findings revealed a correlation between elevated CCDC58 expression and poorer overall survival rate across various factors: Age (Fig. 4D), Gender (Fig. 4E), Histological type (Fig. 4F), 1p/19q codeletion (Fig. 4G), IDH status (Fig. 4H), Race (Fig. 4I), and WHO grade (Fig. 4J).
3.6 Correlation between CCDC58 expression and clinicopathological features in glioma patients
We examined the RNA sequencing data and clinical characteristics of 689 glioblastoma patients in the TCGA database in order to assess the correlation between CCDC58 expression and the clinical pathological aspects of glioma patients. Higher levels of CCDC58 expression were found in Age (> 60 years) (Fig. 5A), Race (Asian) (Fig. 5C), IDH status (WT) (Fig. 5D), 1p/19q codeletion (Codel) (Fig. 5E), WHO grade (G3/G4) (Fig. 5F), and Histological type (Glioblastoma) (Fig. 5H). However, no statistically significant association was observed between CCDC58 expression levels and other clinicopathological features, including Gender (Fig. 5B) and Primary therapy outcome (Fig. 5G). According to these findings, there is a strong correlation between high expression of CCDC58 and the following variables: Race (Asian), IDH status (WT), 1p/19q codeletion (Codel), WHO grade (G3/G4), and Histological type (glioblastoma), respectively.
3.7 Expression of CCDC58 correlated with TMB, MSI, CNV, and epigenetic modifications in different types of cancer
We plotted the mutations of CCDC58 in 12 cancers (Fig. 6A). The analysis indicated that missense mutations were the most common type of genetic alterations in CCDC58, with DLBC (2.7%), UCEC (1.1%), PAAD (0.6%), and ESCA (0.6%) showing the highest frequency of occurrences. Additionally, "shift deletion mutations" and "splice sites" were primarily observed in COAD (1.1%) and COADREAD (0.8%). Furthermore, we investigated the relationship between CCDC58 expression and CNV status across 22 different cancers, identifying statistically significant differences in CCDC58 expression in 15 of these cancers (Fig. 6B), including GBM, LGG, CESC, LUAD, BRCA, STES, SARC, PRAD, UCEC, HNSC, LUSC, OV, PCPG, SKCM, and BLCA. Significantly, frequent gains in the genome were observed for CCDC58, influencing its copy number in gliomas: GBM (loss = 6, gain = 143) and LGG (loss = 8, gain = 499). Additionally, CCDC58 expression was found to be associated with Tumor Mutational Burden (TMB) across 36 cancers (Fig. 6C), showing a positive correlation with TMB in LUSC, HNSC, BRCA, SARC, LGG, STES, BLCA, PRAD, STAD, LUAD, and PAAD. Furthermore, in seven cancers, including STAD, STES, SARC, HNSC, LUSC, LIHC, and PRAD, CCDC58 expression was positively correlated with Microsatellite Instability (MSI) (Fig. 6D), while exhibiting a negative correlation with COAD, OV, COADREAD, and BRCA. Moreover, the analysis of the association between CCDC58 expression and genes related to RNA modification (Fig. 6E) revealed a general positive correlation between CCDC58 expression and the expression of m1A, m5C, and m6A-related genes in glioma, particularly TRMT10C, TRMT61B, TRMT6, ALKBH1, DNMT1, and ALYREF. These findings suggest a close association between the dysregulated expression of CCDC58 in various tumors, its genetic variations, and involvement in epigenetic modifications.
3.8 Correlation between CCDC58 expression and TME in different types of cancer
We assessed the relationship between CCDC58 expression and the tumor microenvironment (TME) in our pan-cancer analysis. Our findings revealed an inverse association between CCDC58 expression and the mesenchymal score in the majority of tumors (Fig. 7). We only did not detect a significant difference in a small number of tumors (ACC, DLBC, KICH, and PCPG, for example) (Figure S5A). A method employed for evaluating the tumor immune microenvironment is immunoscoring. Most tumors exhibited a negative correlation between CCDC58 expression and the immunity score (Figure S4), while a small subset of tumors (ACC, CHOL, DLBC, MESO, PCPG, SARC, and LGG) showed no noticeable difference (Figure S5B). Figure S6A, B illustrates the six tumors with the strongest correlation coefficients between TME and CCDC58 expression.
3.9 Correlation among immune cell infiltration and CCDC58 expression in various cancer types
At last, from 9,405 tumor samples representing 35 different tumor types, we were able to identify six immune cell infiltration categories. We also demonstrated that in 31 malignancies, the expression of the CCDC58 gene was substantially correlated with immune cell infiltration (Fig. 8A). LUSC, STAD, HNSC, and BRCA showed a noteworthy negative correlation between the expression of CCDC58 and six different types of immune cell infiltration. We then looked into the connection between immune cell subtype invasion and CCDC58 expression in 38 tumors. The findings revealed a significant correlation between the extent of immune cell infiltration and CCDC58 expression in most cancer types (37/38) (Fig. 8B). These results indicate an association between CCDC58 expression and the degree of immune cell infiltration across different tumor types.
3.10 CCDC58 promotes glioma cell proliferation, migration, and invasion in vitro
We successfully knocked down the expression of CCDC58 by transfecting LN229 and U251 cell lines with siRNA in order to further confirm the role of CCDC58 in gliomas (Fig. 9A). As a next step we knocked down the gene using siCCDC58#2. The findings of the CCK-8 、wound scratching test and Transwell assay anddemonstrated that down-regulation of CCDC58 hindered migration and invasion as well as cell proliferation in LN229 and U251 (Fig. 9C-H). These findings imply that CCDC58 stimulates glioma cell invasion, migration, and proliferation in vitro.