Molecular Characterization and Clinical Relevance of N6-Methyladenosine Regulators in Metastatic Prostate Cancer

Abstract

Prostate cancer is a leading malignancy in the male population globally. N6-methylation of adenosine (m6A) is the most prevalent mRNA modification and plays an essential role in various biological processes in vivo. However, the potential roles of m6A in metastatic prostate cancer are largely unknown. In this study, we evaluated and identified two m6A modification patterns based on 21 m6A regulators in four public metastatic prostate cancer datasets. Different modification patterns correlated with distinct molecular characteristics. According to m6A-associated genes, we constructed a prognostic model, called m6Ascore, to predict the outcomes of patients with metastatic prostate cancer. We found that high m6A score level was related to dismal prognosis and characterized by higher cell cycle, DNA repair and mismatch repair pathway score. In vitro experiments confirmed that upregulation of METTL14, an m6A writer, enhanced the invasion, metastasis, and sensitivity of prostate cancer cells to poly (ADP-ribose) polymerase inhibitor. Conversely, down-regulation of potential target genes of m6A had the opposite effect. Finally, we validated that a higher m6A score was associated with a worse prognosis and a higher Gleason score in The Cancer Genome Atlas Program (TCGA) cohort. This work illustrated the nonnegligible role of m6A modification in multiple biological processes of metastatic prostate cancer. Evaluating the m6A risk scores of individual tumours will guide more effective judgement of prognosis as well as treatments for metastatic prostate cancer in clinical practice.

Keywords: m6A; metastatic prostate cancer; prognosis; regulator; treatment.

Figure 1

Genetic variants of m6A regulators. (A) The expression of m6A regulator genes in nonmetastatic and metastatic prostate cancers; Frequency of CNV in m6A regulator genes in primary tumour (B), metastatic tumour (C), neuroendocrine prostate cancer (D) and prostate adenoma (E) were shown. Blue represents deletion, orange represents amplification. (F, G) The location of somatic mutations of m6A regulator genes in (F) primary tumour and (G) metastatic tumour. (H) The location of m6A regulator genes on chromosomes. ns represents P > 0.05, *P ≤ 0.05, **P ≤ 0.01.

Figure 2

Unsupervised clustering of m6A regulator genes in metastatic prostate cancer. (A) The interaction among m6A regulator genes. The size of circle indicates the effect of each gene on survival, the larger the size, the greater the effect is; green spots inside the circle indicate risk prognostic factors, black spots inside the circle indicate factors; lines that connect genes exhibit genetic interactions, red and blue represent positive and negative associations, respectively; gene Cluster A, B and C are shown as blue, red and brown, respectively; (B) Consensus clustering m6A regulator genes in metastatic samples; (C) Kaplan–Meier curves indicate that there is a strong relationship between the m6Acluster types and the overall survival rate; (D) GSVA enrichment analysis. Heatmaps show the activation status of biological pathways, which is displayed with different m6A clusters; red denotes activation, blue denotes inhibition; (E, F) show the distribution of the mutation and expression of partial genes in two m6A clusters, respectively.

Figure 3

Comparison analysis among m6A clusters. (A) The distribution of ARV7 (left), ARscore (middle) and NEPCscore (right) between the two m6A clusters; (B) The prognostic differences between the two m6A clusters; (C) Results for GSVA analysis of prad_su2c_2019 cohorts; (D) The expression of m6A regulator genes in two m6A clusters extracted from prad_su2c_2019 cohorts. ns represents P > 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Figure 4

METTL14 promotes PC3 cell metastasis and proliferation in vitro. (A) Western blot analysis of METTL14 expression levels in METTL14-downregulated, METTL14-knockdown, and vehicle control cells. (B) Representative images of migration (upper panels) and invasion (lower panels) assays using PC3 cells, presenting cell migration and invasion after overexpression or knockdown of METTL14. (C) Wound healing assays using PC3 cells presenting cell motility after overexpression or knockdown of METTL14. (D) Cell proliferation was evaluated in METTL14-overexpressing (left) or METTL14-knockdown (right) PC3 cells with or without olaparib administration. ****P ≤ 0.0001.

Figure 5

Construction of m6A risk score model. (A) The alluvial plot shows the changes of m6A clusters, gene clusters and m6Ascore; (B) Kaplan–Meier curves indicate that there is a strong relationship between the m6Ascore and the overall survival rate; (C) Pearson’s correlation analysis highlighting the correlations between m6Ascore and the known gene ontologies in prad_su2c_2019 cohorts. Red, blue and X symbols represent positive, negative and nonsignificant, respectively; the larger the circle, the more significant there is; (D) The distribution of enrichment scores of known gene ontologies prad_su2c_2019 cohorts between high and low m6Ascore samples; (E, F) show the distribution of m6Ascore among m6Aclusters and m6Ageneclusters, respectively. ns represents P > 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Figure 6

CSNK1D or SLC35E1 ablation promotes PC3 cell metastasis and proliferation in vitro. (A) Western blot analysis of CSNK1D or SLC35E1 expression levels in CSNK1D or SLC35E1 knockdown cells and vehicle control cells. (B) Representative images of migration (upper panels) and invasion (lower panels) assays using PC3 cells, presenting cell migration and invasion after knockdown of CSNK1D or SLC35E1. (C) Wound healing assays using PC3 cells presenting cell motility after knockdown of CSNK1D or SLC35E1 ablation. (D) Cell proliferation was evaluated in CSNK1D (left) or SLC35E1 (right) ablated PC3 cells with or without olaparib administration. **P ≤ 0.01, ****P ≤ 0.0001.

Figure 7

Molecular profiling of sample groups with high and low m6Ascore. (A) The distribution of ARV7 (middle), ARscore (middle) and NEPCscore (right) between samples with high and low m6Ascore; Gene mutation distribution of high (B) and low (C) m6Ascore samples; The distribution of copy number amplifications and deletions in high (D) and low (E) m6Ascore samples. ns, no significance; **P ≤ 0.01.

Figure 8

Comparison analysis and validation of m6Ascore model. (A) Survival analysis plot indicates a significant difference between TCGA samples with high and low m6A score. (B, C) The distribution of m6A score within distinct T stages and GLEASON_SCORE subgroups using TCGA data. (D) ROC curves for the prediction of metastatic and nonmetastatic prostate cancer between groups with high and low m6A score.