PO.MCB10.01 · 分子与细胞生物学
通过计算共识靶向和功能聚类对NSCLC中亚型感知的miRNA网络进行图谱绘制
Isoform-aware miRNA network mapping in NSCLC via computational consensus targeting and functional clustering
作者与单位 Authors & Affiliations
摘要 Abstract
中文摘要
由可变加工、RNA编辑和加尾/修剪产生的microRNA（miRNA）异构体（isomiR）拓宽了调控范围，并可能在靶点和功能上有别于经典miRNA。然而，大多数癌症基因组学工作流程仍将测序读段归并为经典miRNA或忽略非经典变体，从而掩盖了异构体特异性调控网络。这一缺口在NSCLC中尤为相关，该病异质性高，且在生物标志物和靶点方面存在高度未满足的需求。基于我们的泛癌数据显示isomiR感知分析可改善临床病理分类，我们假设系统性的异构体水平分析将揭示传统miRNA分析所遗漏的NSCLC特异性生物学特征和易感性。
来自TCGA LUAD和LUSC的数据以异构体感知流程进行处理。使用|倍数变化|>1.5和FDR<0.05定义肿瘤与配对正常组织之间的差异表达isomiR。在六种算法间预测共识靶点，保留由≥4种工具支持且在isomiR高表达相对于isomiR低表达肿瘤中下调的相互作用。使用通路水平靶点矩阵上的Jaccard距离、层次聚类以及跨肺癌相关通路的通路频率汇总，比较各亚型间的异构体功能。
我们在LUAD中鉴定出约1,300个失调isomiR，在LUSC中约1,100个。基于通路的聚类将isomiR分组为区分LUAD和LUSC的功能簇。富集于标志性致癌信号网络（如PI3K-AKT、MAPK、p53、VEGF）的簇包含约14%汇聚于这些通路的共享isomiR。贡献最大异构体谱系的miR-183-5p与高簇活性以及PTEN/p53-AKT信号一致。致癌簇进一步由miR-17-5p/miR-93-5p和情境依赖的miR-130b-3p塑造，而抑癌isomiR（如miR-128-3p）则标志着对细胞周期进展和侵袭的抑制。
异构体感知的miRNA分析揭示了超越传统视角的NSCLC调控网络。将共识靶向与队列水平表达和基于通路的聚类整合，能够对isomiR进行功能鉴定，并对异构体-靶点对（包括亚型特异性和共享致癌网络）进行优先排序，作为潜在的生物标志物和治疗靶点。这些结果仍受限于isomiR定量、情境特异性靶点预测、队列变异性等当前挑战以及功能验证的需求。
查看英文原文 English abstract
MicroRNA (miRNA) isoforms (isomiRs), generated by alternative processing, RNA editing, and tailing/trimming, broaden regulation and can differ from canonical miRNAs in targets and function. Yet most cancer genomics workflows still collapse reads to canonical miRNAs or ignore non-canonical variants, obscuring isoform-specific regulatory networks. This gap is especially relevant in NSCLC, where heterogeneity and unmet needs in biomarkers and targets are high. Building on our pan-cancer data showing that isomiR-aware profiling improves clinicopathologic classification, we assumed that systematic isoform-level analysis would uncover NSCLC-specific biological signatures and vulnerabilities missed by traditional miRNA analyses.
Data from TCGA LUAD and LUSC were processed with an isoform-aware pipeline. Differentially expressed isomiRs between tumors and matched normals were defined using |fold-change|>1.5 and FDR<0.05. Consensus targets were predicted across six algorithms, retaining interactions supported by ≥4 tools and downregulated in isomiR-high versus isomiR-low tumors. Isoform functions were compared across subtypes using Jaccard distances on pathway-level target matrices, hierarchical clustering, and pathway-frequency summaries across lung cancer-related pathways.
We identified ~1,300 deregulated isomiRs in LUAD and ~1,100 in LUSC. Pathway-based clustering grouped isomiRs into functional clusters that separated LUAD and LUSC. Clusters enriched for hallmark oncogenic signaling networks (e.g., PI3K-AKT, MAPK, p53, VEGF) contained ~14% shared isomiRs that converged on these pathways. miR-183-5p, which contributed the largest isoform repertoire, aligned with high cluster activity and PTEN/p53-AKT signaling. Oncogenic clusters were further shaped by miR-17-5p/miR-93-5p and context-dependent miR-130b-3p, whereas tumor-suppressive isomiRs (e.g., miR-128-3p) marked restraint of cell-cycle progression and invasion.
Isoform-aware miRNA analysis uncovers NSCLC regulatory networks beyond traditional views. Integrating consensus targeting with cohort-level expression and pathway-based clustering enables functional identification of isomiRs and prioritization of isoform-target pairs, including subtype-specific and shared oncogenic networks, as potential biomarkers and therapeutic targets. These results are still limited by current challenges in isomiR quantification, context-specific target prediction, cohort variability, and the need for functional validation.
利益披露 Disclosure
S. Gu, None..
J. Liu, None..
S. Feng, None..
R. Distefano, None..
S. Di Bella, None..
F. Orilio, None..
R. Brancaccio, None..
G. Romano, None..
E. Shankar, None..
M. Acunzo, None..
F. Calore, None..
C. Rolfo, None..
Q. Ma, None..
G. Nigita, None.