* 일시 : 3/19(목) 16:00 –
* 연사 : 한지연 박사 (NCI-Designated Cancer Center, Sanford Burnham Prebys Medical Discovery Institute)
* 장소 : 벤처관 711호
* 제목 : Widespread and Functional Extreme Gene Expression in Cancer
*초록 :
Cancer evolves under diverse selective pressures and is characterized by widespread dysregulation of gene expression. Transcriptomic studies have traditionally relied on differential expression analyses between predefined patient groups to identify genes associated with specific clinical or molecular phenotypes. However, cancers also exhibit rare, extreme gene expression states (XEGs) that appear as outliers relative to a population of cells or individuals. Using a novel outlier detection algorithm, OutSeekR, we performed a comprehensive analysis of extreme gene expression across 7,019 breast tumours from multiple independent cohorts. XEGs were unexpectedly common, with approximately three quarters of tumours harbouring at least one XEG. The frequency of XEGs varied substantially by molecular subtype, with more aggressive subtypes exhibiting significantly higher numbers of XEGs per tumour compared to less aggressive subtypes. These differences were associated with patient survival within specific molecular subtypes. XEGs were observed across multiple layers of the central dogma. RNA XEGs were reflected at the protein level in matched mass spectrometry and reverse phase protein array data, and approximately half of XEGs were associated with underlying somatic mutations, copy number alterations, or aberrant promoter DNA methylation. XEGs identified in patient tumours were also recapitulated in cancer cell lines, enabling functional interrogation. Genetic perturbation of XEG-associated genes using CRISPR or RNA interference, as well as pharmacologic inhibition, selectively reduced cancer cell fitness in cell lines harbouring the corresponding XEG, but not in those without. Together, these results demonstrate that extreme gene expression is common in breast cancer, can confer a selective advantage to tumour cells, and may represent a previously underappreciated class of targetable vulnerabilities.