Construction and validation of a novel cuproptosis-mitochondrion prognostic model related with tumor immunity in osteosarcoma

Abstract

Background: The purpose of this study was to develop a new prognostic model for osteosarcoma based on cuproptosis-mitochondrion genes.

Materials and methods: The data of osteosarcoma were obtained from TARGET database. By using Cox regression and LASSO regression analysis, a novel risk score was constructed based on cuproptosis-mitochondrion genes. Kaplan-Meier, ROC curve and independent prognostic analyses were performed to validate the risk score in GSE21257 dataset. Then, a predictive nomogram was constructed and further validated by calibration plot, C-index and ROC curve. Based on the risk score, all patients were divided into high-risk and low-risk group. GO and KEGG enrichment, immune correlation and drug sensitivity analyses were performed between groups. Real-time quantitative PCR verified the expression of cuproptosis-mitochondrion prognostic model genes in osteosarcoma. And we explored the function of FDX1 in osteosarcoma by western blotting, CCK8, colony formation assay, wound healing assay and transwell assays.

Results: A total of six cuproptosis-mitochondrion genes (FDX1, COX11, MFN2, TOMM20, NDUFB9 and ATP6V1E1) were identified. A novel risk score and associated prognostic nomogram were constructed with high clinical application value. Strong differences in function enrichment and tumor immune microenvironment were shown between groups. Besides, the correlation of cuproptosis-mitochondrion genes and drug sensitivity were revealed to search for potential therapeutic target. The expression of FDX1, COX11, MFN2, TOMM20 and NDUFB9 at mRNA level was elevated in osteosarcoma cells compared with normal osteoblast hFOB1.19. The mRNA expression level of ATP6V1E1 was decreased in osteosarcoma. Compared with hFOB1.19, western blotting revealed that the expression of FDX1 was significantly elevated in osteosarcoma cells. Functional experiments indicated that FDX1 mainly promoted the migration of osteosarcoma rather than proliferation.

Conclusions: We developed a novel prognostic model of osteosarcoma based on cuproptosis-mitochondrion genes, which provided great guidance in survival prediction and individualized treatment decision making for patients with osteosarcoma.

Fig 1. The selection of cuproptosis-mitochondrion genes for further construction of risk score.

(A) Protein-protein interaction (PPI) network of the cuproptosis-mitochondrion genes. (B) Correlation network of cuproptosis-mitochondrion genes. (C) Six cuproptosis-mitochondrion genes were associated with prognosis of osteosarcoma in univariate Cox regression analysis. (D and E) After LASSO regression analysis, all of the six cuproptosis-mitochondrion genes were selected to construct risk score.

Fig 2

The construction and validation of the prognostic risk score based on six cuproptosis-mitochondrion genes (A, C, E, G and I for construction cohort from TARGET database; B, D, F, H and J for validation cohort from GSE21257). The Kaplan-Meier curve of patients in high-risk and low-risk group (A and B). The ROC curve of the 1-, 3- and 5- year overall survival (C and D). The distribution of risk score for each osteosarcoma patient (E and F). The overall survival rate and its survival status of osteosarcoma patients (G and H). The expression of these six cuproptosis-mitochondrion genes in the low-risk group and the high-risk group. Cool color represents low expression while warm color represents high expression (I and J).

Fig 3. Independent prognostic analysis and the nomogram predicting the survival outcome of osteosarcoma.

The univariate (A) and multivariate Cox (B) regression analyses were performed. (C) The nomogram was constructed based on gender, age, metastasis and risk score. (D) The calibration curve for predicting 1-,3‐, and 5-year overall survival. (E) The C-index of predictive factors in the nomogram according to times. The 1-year (F), 3-year (G), and 5-year (H) ROC curves of predictive factors are used to calculate AUC values.

Fig 4. Gene set enrichment analysis between the low-risk group and high-risk group.

(A) Enriched GO terms between high-risk group and low-risk group. (B) Enriched KEGG pathways between high-risk and low-risk group.

Fig 5. Immune correlation analysis.

(A) The bar plot of 22 immune cell infiltrations in CIBERSORT. (B) Violin plot showed the ratio differentiation of 22 kinds of immune cells between osteosarcoma samples with low or high expression of FDX1, and Wilcoxon rank sum was used for the significance test. (C-E) T cells gamma delta and Dendritic cells resting were positively connected with the expression of FDX1 while CD8+ T cells was negatively correlated with the expression.

Fig 6. The expression of immune checkpoint genes between low-risk and high-risk group and its correlation with risk score.

(A-H) The comparison of the expression levels of LAG3, HAVCR2, CD27, TNFRSF14, CTLA4, TMIGD2, TIGIT and PDCD1LG2 between high-risk and low-risk group. (I-P) Significant association between immune checkpoint genes and risk score were identified.

Fig 7. The correlation between six cuproptosis-mitochondrion genes and sensitivity of drug.

Gene expression is depicted horizontally while drug sensitivity is depicted vertically. Correlation coefficient R>0 was considered as a positive correlation, and P<0.05 was considered as a significant difference.

Fig 8. The mRNA expression level of cuproptosis-mitochondrion prognostic model genes.

*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 each experiment was repeated three times.

Fig 9. The expression and migration of FDX1 in osteosarcoma.

(A)The protein expression levels of FDX1 in hFOB1.19, U2OS, MG63, and 143B. The migration ability of U2OS cell in Vector-NC group and FDX1-OE group by transwell assay without Matrigel (B) and wound healing assay (C). each experiment was repeated three times. *P<0.05, **P<0.01, ***P<0.001*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 each experiment was repeated three times.