DZIP3 is associated with the cytoskeletal dynamics in glioblastoma and mitochondrial machinery regulation

Abstract

Glioblastoma (GBM) is a major public health issue with increased prevalence and incidence worldwide. The therapeutic strategy is based on surgical resection, followed by chemotherapy and radiotherapy. However, therapeutic failure is increasingly recognized as driven by the intricate characteristics of the tumor microenvironment (TME). Stress granules (SGs) are structures that emerge in cells in response to stress. In GBM, these elements are implicated in tumor development and progression. Heterogeneous elements coalesce to form SGs, with RNA-binding proteins (RBPs) among them. DZIP3, DAZ-interacting zinc finger 3, is an understudied RBP in cancer, particularly with respect to its impact on TME remodeling. In the current study, we evaluate the prognostic value of DZIP3 using The Cancer Genome Atlas and the Chinese Glioma Genome Atlas as the validation cohort. Initially, high DZIP3 expression was associated with better survival in the TCGA cohort, but after adjustment for age, gender, transcription subtypes, IDH mutation status, MGMT promoter methylation status, and Karnofsky performance score, the prognostic utility of DZIP3 did not persist. Similar results were observed in the CGGA cohort. To identify pathways enriched among upregulated genes associated with high DZIP3 expression, we performed multiple enrichment analyses. To evaluate whether the gene sets obtained are indeed associated with DZIP3, we conducted a robustness assessment and retained only those that met the pre-specified FDR and effect-size thresholds. Of the 30 modules obtained, 7 retained independent association: the cytoskeletal dynamics/intracellular transport gene set was the strongest correlate of DZIP3, followed by 4 mitochondrial modules (branched-chain amino acid metabolism, NAD biosynthesis, translation factors, and pyruvate metabolism), RNA granule/neurodegeneration gene set, and the nucleocytoplasmic transport gene set. The immune checkpoints, P-bodies, and malignant cell state gene sets were not independent of this signal. The cytoskeletal signal was replicated in the CGGA cohort and at single-cell resolution, with 9 core hub miRNAs targeting DZIP3, suspected to be a regulatory mechanism. Altogether, this study identifies the cytoskeletal module, 4 mitochondrial programs, RNA granule/neurodegeneration, and nucleocytoplasmic transport gene sets as the strongest correlates of DZIP3 expression in GBM, demonstrating that clinical prognostic value is distinct from transcriptional reproducibility.