The upregulated expression of mitocytosis-related genes in several cancer
We analyzed the mRNA expression levels of MYO19, KIF5B, DNM1L, DYNLL2, TSPAN4, and TSPAN9 using data from TCGA. The analysis revealed that, compared to adjacent normal tissues (including UCEC and KICH), the expression of these genes is generally lower in some tumor types. However, certain cancers such as BRCA, CHOL, COAD, ESCA, HNSC, LIHC, and STAD exhibited a significant upregulation trend (Fig. 1A). Since TSPAN4 and DYNLL2 are not included in the HPA database, we used IHC results from the HPA database to confirm the protein-level expression of MYO19, KIF5B, DNM1L, and TSPAN9 (Fig. 1B).
IHC staining demonstrated that these genes exhibited weak or negative expression in tumor tissues from gastric cancer and renal cancer. Overall, we demonstrated decreased expression of MYO19, KIF5B, DNM1L, DYNLL2, and TSPAN9 in these tumors.
The prognostic value of mitocytosis-related genes in certain types of cancer.
Using raw sample data from the TCGA database, we investigated the relationship between the expression levels of MYO19, KIF5B, and DNM1L and overall survival (OS) in 33 cancers by univariate Cox regression analysis (Fig. 2A). MYO19 showed significant hazard ratios in LGG, MESO, PRAD, READ, and UVM. KIF5B showed significant association in ACC, KIRC, THYM, and UVM. DNM1L showed significant hazard ratios in ACC, BRCA, LUAD, MESO, and READ. Additionally, we evaluated the prognostic relevance of mitocytosis-related genes in cancer patients. Upregulated expression of MYO19, KIF5B, DNM1L, DYNLL2, TSPAN4, and TSPAN9 was associated with lower OS in ACC, LIHC, and MESO, but higher OS in CRC, KIRC, and SKCM (Figs. 2B and S-1). ROC curve analysis indicated that MYO19 (AUC = 0.845) and DNM1L (AUC = 0.842) may act as diagnostic markers, as demonstrated in HNSC. We also analyzed the expression levels of MYO19, KIF5B, and DNM1L in LIHC, MESO, and READ across different ages, genders, and WHO stages, revealing significant differences between groups (Figure S-2).
The positive correlation between mitcytosis-related gene CNV, SNV and methylation
Pearson correlation analysis showed a strong correlation between CNV and the mRNA RSEM of DNM1L, DYNLL2, and TSPAN9 in TCGA (Fig. 3A). However, BRCA and READ showed a poor correlation for KIF5B, MYO19, and TSPAN9. CNV patterns varied by gene and cancer type, with heterozygous amplification and deletion being the main types. TSPAN9 and DNM1L had low frequencies of heterozygous deletion and high frequencies of heterozygous amplification. Homozygous amplification was observed for TSPAN9 and DNM1L in OV and TGCT, and for MYO19 and TSPAN9 in CHOL, while TSPAN4 had homozygous deletions.
In methylation analysis, we compared tumor and normal tissues, revealing significant disparities. MYO19 showed low methylation levels in BRCA, HNSC, BLCA, LIHC, LUAD, PRAD, and UCEC, while DYNLL2, KIF5B, and DNM1L showed elevated levels in KIRC. Spearman correlation analysis indicated high methylation levels of TSPAN4 in ACC, PRAD, and LIHC, and for DYNLL2, TSPAN4, and TSPAN9 in THYM. When comparing OS between high and low methylation levels, TSPAN4 hypermethylation had a low effect on survival risk in SKCM, GBM, PAAD, and LGG, whereas TSPAN9 hypermethylation had a high effect on survival risk in ACC, LUAD, THCA, and UVM. In mutation analysis of 301 samples, KIF5B had the highest mutation count and frequency across all samples, particularly in UCEC, followed by MYO19 and DNM1L (Fig. 3C).
The innate mechanism unravelled through GSEA
We analyzed the expression levels of 1609 genes in KIRC and 5415 genes in MESO, significant differences were found between the high- and low-expression groups. In KIRC, 63% genes were upregulated, and 37% were downregulated, while in MESO, 66% genes were upregulated, and 34% were downregulated (with a p-value < 0.05, |Log2-FC| > 1) (Fig. 4A). The Enrichment Score results indicate that genes associated with the Intermediate Filament Cytoskeleton are predominantly highly expressed (Fig. 4B). A heatmap further illustrates this trend, showing a significant increase in log fold change (logFC) for these genes, while other genes appear randomly distributed.
Immune-related characteristics of mitocytosis-related gene in pan-cancer
To investigate the role of mitocytosis-related genes in immune mechanisms and immune response within the TME, we explored the potential correlation between the expression of these six genes and immune cell infiltration in human cancers. TMB can indirectly reflect the ability and extent of a tumor to produce new antigens, while microsatellite instability (MSI) is also related to cellular immunogenicity and can predict the efficacy of immunotherapy in various tumors. The results showed that the expression of MYO19 was significantly positively correlated with TMB in ACC, UCS, UCEC, STAD, SKCM, READ, PRAD, LUAD, KIRC, and COAD. KIF5B expression showed a negative correlation with COAD and positive correlations with GBM, LUAD, SKCM, and UCEC. DNM1L expression was negatively correlated with KIRP and THCA, and positively correlated with LUAD, READ, and UCEC. In terms of MSI, MYO19 expression demonstrated a positive correlation with UVM, UCEC, STAD, SARC, PRAD, LUSC, LUAD, LGG, KIRC, KICH, COAD, and BRCA, and a negative correlation with DLBC. KIF5B expression showed a negative correlation with COAD, DLBC, HNSC, PRAD, and THCA, and a positive correlation with CESC and KIRC. DNM1L expression exhibited a negative correlation with DLBC and PRAD, while being positively correlated with CESC, COAD, KIRC, LUSC, READ, STAD, TGCT, and UCEC (Fig. 5A).
In the tumor microenvironment, tumor-infiltrating immune cells (TIICs) are a critical component and influence the prognosis of cancer patients. We used the TIMER database to analyze the relationship between the expression levels of MYO19, KIF5B, DNM1L, DYNLL2, TSPAN4, TSPAN9, and DYNLL2 and immune cell infiltration. We observed a significant correlation between the expression levels of these genes and CD4 + T cell, particularly demonstrating elevated levels collectively in BLCA, HNSC, KIRC, THCA, UCEC, and UVM. We found a significant positive correlation of KIRC between CD4 + T cell level and MYO19, KIF5B, DNM1L, DYNLL2, TSPAN4, TSPAN9 and DYNLL2 expression in MYO19 (Rho = 0.104), TSPAN4(Rho = 0.488) and TSPAN9 (Rho = 0.43) (Fig. 5B and Figure-S-3).
Next, we conducted an in-depth analysis of the correlation between mitocytosis-related gene expression and immune scores across all cancer types (Fig. 5C). The results indicated that MYO19 expression had a significant correlation with the three immune scores, showing particularly strong correlations in GBM, KIRP, LUSC, SARC, and TGCT. KIF5B exhibited a high correlation in ACC, CESC, SARC, STAD, and THYM. DNM1L showed high correlation in ACC, GBM, LGG, MESO, LUSC, SARC, and TGCT, all with significant statistical relevance.
Clinical prognostic value of mitocytosis-related gene
In order to explore the clinical prognostic value of mitocytosis-related genes, this study combined mitocytosis-related gene expression with clinical factors based on time ROC to construct a nomogram and calibration analysis. In the survival-related ROC curves, it is evident that only the 5-year duration exhibits a strong correlation (AUC = 0.709). We analyzed the clinical value by the ROC curve (Fig. 6A). Mitocytosis-related gene perform as a clinical marker in risk score (AUC = 0.709) and stage (AUC = 0.724). Subsequently, we compared the performance of high and low risk scores in survival probability (Fig. 6B), revealing significantly higher overall survival probability associated with low risk scores.
Subsequently, we investigated the interrelationships among these six genes (Fig. 6C). The strongest positive association was shown between KIF5B and DYNLL2, while the strongest negative association was shown between DNM1L and TSPAN4.Next, we combined clinical features with risk score in both multivariable and univariable analyses to construct forest plots (Fig. 6D). In the univariable forest plot, stage, T_stage, N_stage, M_stage, and risk score showed statistical significance. However, in the multivariable forest plot, only risk score demonstrated statistical significance.
The potential influence of mitocytosis-related genes on IPS scores and drug sensitivity.
We performed unsupervised clustering analysis on 529 KIRC samples from TCGA based on the expression levels of mitocytosis-related genes. We applied the "ESTIMATE" and "CIBERSORT" algorithms to determine the expression levels of mitocytosis-related genes in tumor samples. Using the R package "ConsensusClusterPlus," we clustered all samples based on these expression levels, identifying two distinct subtypes (Fig. 7A). The prognosis of these two subtypes was compared using Kaplan-Meier curves, and the results showed that the prognosis of group A was significantly better than that of group B (p = 0.001). Based on these findings, patient samples were classified into high-risk and low-risk prognostic groups (HSG and LSG, respectively). Previous studies have shown that immunogenicity-based IPS can play a role in predicting the efficacy of immunotherapy in KIRC patients. We analyzed the relationship of IPS between HSG and LSG. We used IPS- PD-L1 + CTLA4+, IPS- PD-L1-CTLA4+, IPS- PD-L1-CTLA4-, IPS- PD-L1 + CTLA4- to evaluate the potential of scores application (Fig. 7B). All 4 groups showed significant differences between HSG and LSG (all p < 0.05). Through overlap analysis of KIRC and MESO associated drugs, 22 overlapping drugs were identified by KIRC risk, MESO risk, KIRC 1vs2, MESO 1vs2 (Fig. 7C). Subsequently, we compared the sensitivity of the two clusters to a subset of treatment-related drugs in KIRC and MESO. The study demonstrated the predicted therapeutic responses of anti-tumor drugs commonly used clinically by KIRC and MESO, including telomerase inhibitor IX, vinblastine, and leflunomide, in two clusters of HSG and LSG samples, and significant group differences were observed. Differences between (all p < 0.05)