Identification of miRNA-based biomarkers predictive of lung cancer treatment response and of mechanisms involved in lung cancer progression.

Lung cancer remains the leading cause of cancer-related mortality worldwide, with non-small-cell lung cancer (NSCLC) accounting for nearly 85% of all cases. Despite advances in targeted therapy and immunotherapy, overall survival remains poor due to late-stage diagnosis and the lack of reliable biomarkers predictive of disease progression and treatment response. MicroRNAs (miRNAs) are small, non-coding RNAs of approximately 22 nucleotides which are critical post-transcriptional regulators of gene expression and play central roles in cancer initiation, progression, and therapeutic resistance. However, a comprehensive functional screening of the human miRNome in lung cancer has been lacking. This PhD project aimed to systematically map the human miRNome in NSCLC through a high-throughput functional screen, integrating molecular, phenotypic, and clinical datasets to identify miRNA-based biomarkers and novel therapeutic targets. Using a pooled lentiviral library encompassing 2,580 human miRNAs, lung adenocarcinoma (LUAD) cell lines (e.g., A549) were stably transduced and selected with puromycin to generate a heterogeneous population expressing the human miRNome. These cells were subsequently subjected to independent selective pressures corresponding to key hallmarks of cancer, including immune evasion (using PD-L1 expression as a proxy), resistance to cell death (using combination chemotherapy as a proxy), proliferation, migration, and invasion, to functionally enrich for phenotype-specific miRNAs. Genomic DNA from reference and selected cell populations was extracted, PCR-amplified, and analysed by next-generation sequencing (NGS) to retrieve the relative abundance of miRNA constructs. Differential enrichment and depletion analyses using DESeq2 and QTrait identified distinct subsets of miRNAs driving each phenotype, with minimal overlap across proliferation, migration, and invasion. Integration of functional screening data with transcriptomic and clinical information from The Cancer Genome Atlas Lung Adenocarcinoma (TCGA-LUAD) cohort, as well as in-house spatial transcriptomics datasets, led to the identification of a 15-miRNA prognostic signature strongly associated with poor overall survival (Hazard Ratio = 2.48, p < 0.0001). Among these, miR-92b-3p, an understudied member of the oncogenic miR-17~92 cluster, emerged as a potent driver of lung cancer aggressiveness. Mechanistic investigations revealed that miR-92b-3p promotes migration and invasion through activation of NOTCH3 signalling. Transcriptomic profiling confirmed upregulation of NOTCH3 and its downstream targets upon miR-92b-3p overexpression. Conversely, both genetic and pharmacologic inhibition of NOTCH3 completely abrogated the pro-migratory and invasive effects of miR-92b-3p, establishing a novel miR-92b-3p/NOTCH3 axis as a key determinant of LUAD progression. In summary, this study provides a systematic functional annotation of the human miRNome in lung adenocarcinoma, revealing novel regulators of tumour aggressiveness and metastatic behaviour. The identification of a robust 15-miRNA prognostic signature and the characterization of the miR-92b-3p/NOTCH3 signalling axis significantly advance our understanding of miRNA-mediated mechanisms involved in NSCLC progression. These findings demonstrate that miRNAs serve both as clinically relevant biomarkers for patient stratification and as promising therapeutic targets for reversing resistance and improving treatment outcomes in lung cancer.