Multi-omics comprehensive analysis identified KIF22 and KRAS as highly synthetic lethal pairs for triple-negative...

Triple-negative breast cancer (TNBC) is an aggressive subtype of breast cancer with a poor prognosis and limited treatment options. Synthetic lethality (SL) represents a significant therapeutic strategy that selectively kills cancer cells without affecting normal cells by targeting the synergistic interaction of two genes. The SL strategy offers new avenues for targeted therapy in TNBC. Although challenges remain-such as drug resistance and biomarker selection-advancing research in SL activity holds promise for delivering clinical benefits to patients.

Multi-omics data, including single-cell RNA sequencing (scRNA-seq), spatial transcriptomics (ST), and bulk RNA sequencing (bulkRNA-seq), were utilized to characterize TNBC heterogeneity and identify genes driving SL activity. Additionally, CytoTRACE analysis assessed tumor differentiation potential in high SL (HSL) activity cell subpopulations, Slingshot reconstructed pseudo-temporal trajectories, and CellChat constructed intercellular communication networks to evaluate interactions among TNBC microenvironment cell subpopulations. Combining high-dimensional weighted gene co-expression network analysis (hdWGCNA) with machine learning, key regulatory genes associated with the HSL cell phenotype were identified. Finally, a benchmarking framework was employed to select the most predictive algorithmic model, with feature contributions evaluated via SHapley Additive Explanations (SHAP) analysis. The identified genes were analyzed in vivo and in vitro through molecular biology experiments and animal experiments.

A novel HSL subtype of TNBC malignant cells has been identified, exhibiting enhanced stem cell-like properties, stronger intercellular communication capabilities, and involvement in more tumor-associated signaling pathways. Ten characteristic genes identified through five machine learning (PGP, KIF22, CCNB1, RPA3, BCL2L12, SMC2, MKI67, PBK, CDK1, and MIS18A) are significantly upregulated in TNBC malignant cells, and their high expression correlates with poor prognosis in TNBC patients. Benchmarking validated the superior performance of the random forest algorithm. Finally, through experimental verification, it was concluded that KIF22 and KRAS are synthetic lethal pairs for TNBC.

In conclusion, this study systematically characterized the heterogeneity of TNBC and explored the association between SL activity and disease progression through a comprehensive analysis of the interactions between SL pairs and malignant TNBC cells. Our findings contribute to a deeper understanding of the molecular mechanisms underlying TNBC initiation and development. Based on bioinformatics analyses, we experimentally validated KIF22 and KRAS as a synthetic lethal gene pair in TNBC. Functional experiments demonstrated that the knockdown of KIF22 in KRAS-mutated TNBC cells or the knockdown of KRAS in TNBC cells with low expression of KIF22 gene significantly inhibited cell proliferation. Given the high prevalence of KRAS mutations in TNBC, KIF22 represents a promising therapeutic target for synthetic lethal intervention. Furthermore, in vivo xenograft models confirmed that concurrent knockdown of murine KIF22 and KRAS effectively inhibited tumor progression. Collectively, these results establish KIF22 and KRAS as a TNBC-specific synthetic lethal pair with strong potential for guiding future SL-based drug discovery efforts.