Abstract
Esophageal squamous cell carcinoma (ESCC) demonstrates high genome instability. Here, we analyze 528 whole genomes to investigate structural variations' mechanisms and biological functions. SVs show multi-mode distributions in size, indicating distinct mutational processes. We develop a tool and define five types of complex rearrangements with templated insertions. We highlight a type of fold-back inversion, which is associated with poor outcomes. Distinct rearrangement signatures demonstrate variable genomic metrics such as replicating time, spatial proximity, and chromatin accessibility. Specifically, fold-back inversion tends to occur near the centrosome; TD-c2 (Tandem duplication-cluster2) is significantly enriched in chromatin-accessibility and early-replication region compared to other signatures. Analyses of TD-c2 signature reveal 9 TD hotspots, of which we identify a hotspot consisting of a super-enhancer of PTHLH. We confirm the oncogenic effect of the PTHLH gene and its interaction with enhancers through functional experiments. Finally, extrachromosomal circular DNAs (ecDNAs) are present in 14% of ESCCs and have strong selective advantages to driver genes.