Abstract
MicroRNA (miRNA) precursors produce two arms, miR-5p and miR-3p, which were traditionally thought to consist of one functional strand and one degraded "passenger" strand. This study challenges that view by demonstrating that both arms can be concordantly dysregulated in lung cancer, working together to regulate protein-protein interaction (PPI) networks. Using a novel computational framework incorporating second-order partial correlation (SOPC), researchers identified 17 high-confidence miR-5p/-3p pairs in lung cancer that target 35 protein complexes. Notably, the down-regulation of the miR-145-5p/-3p pair was found to activate the FOXM1 transcription factor network, driving uncontrolled cell proliferation. This research provides a new lens through which to view miRNA-mediated gene regulation in cancer.
Introduction
MicroRNAs (miRNAs) are critical post-transcriptional regulators of gene expression. The canonical model of miRNA biogenesis suggests that while two strands (5p and 3p) are produced from a single precursor, only one is selected for functional activity while the other is discarded as cellular junk. However, emerging evidence suggests that both arms can be functional. This study addresses how "concordantly dysregulated" miR-5p/-3p pairs—where both arms are simultaneously up- or down-regulated—coregulate biological networks to promote tumorigenesis, specifically within lung cancer.
Methodology and Data Integration
The researchers developed a robust computational framework to analyze matched miRNA and mRNA expression profiles from The Cancer Genome Atlas (TCGA), focusing on 279 lung squamous cell carcinoma (LUSC) samples.
As illustrated in the workflow (Figure 5), the process involves several key steps:
- Differential Expression Analysis: Identifying miRNA pairs where both 5p and 3p arms are significantly dysregulated in tumor versus normal tissues.
- Network Mapping: Integrating these miRNAs with "Coexpressed Protein-Protein Interaction" (CePPI) networks to identify target genes that physically interact and are co-expressed.
- Second-Order Partial Correlation (SOPC): A critical innovation to minimize false positives. SOPC measures the correlation between two target genes after mathematically accounting for the influence of both miRNA arms. A significant drop in correlation compared to standard correlation suggests the miRNAs are the primary drivers of the genes' co-expression.
Results: Concordant Dysregulation and Network Tuning
The analysis revealed that concordant dysregulation is a widespread phenomenon across multiple cancer types. In lung cancer, 23 miRNA pairs showed significant concordant changes, meaning their expression levels moved in the same direction (either both up or both down).
As shown in Figure 1, many miRNA arms move in tandem (represented by points in the top-right and bottom-left quadrants). After applying the SOPC filter, the team narrowed this down to 17 high-confidence pairs. These pairs do not just target individual genes but appear to act as "network tuners" for entire protein complexes. Specifically, 35 complexes involved in cell cycle checkpoints, apoptosis, and differentiation were identified as key targets.
The miR-145/FOXM1 Axis and Clinical Implications
A primary finding of the study is the role of the miR-145-5p/-3p pair. In lung cancer, both arms are significantly down-regulated. This "double loss" of molecular brakes leads to the over-expression of the FOXM1 transcription factor network, a well-known driver of the cell cycle. Validation using Reverse-Phase Protein Array (RPPA) data from 104 samples confirmed that low miR-145 levels correlate with high protein levels of FOXM1 and its downstream effectors, CCNB1 and CCNE2. This confirms the regulatory relationship translates to actual protein changes that drive tumor growth.
Conclusion
This study highlights the importance of considering both arms of a miRNA precursor in cancer research. By moving beyond the "one functional strand" dogma, the authors have uncovered a more complex layer of gene regulation. The developed SOPC-based framework offers a powerful tool for identifying true coregulatory relationships in large-scale genomic datasets, paving the way for identifying new therapeutic targets that address the synergistic effects of miRNA pairs.
Summary updated