The burden of abundance: Oncogene-independent vulnerabilities of cells with extrachromosomal DNA

Cancer genomes are often extensively rearranged, altering the normal organization of genetic material. One striking consequence of this genomic instability is the formation of extrachromosomal DNA (ecDNA), circular DNA molecules that exist outside the chromosomes and are frequently found in cancer cells. ecDNA often carries amplified oncogenes, increasing their copy number and expression, and has been associated with tumor progression and therapy resistance. Unlike chromosomes, ecDNA lacks centromeres, can undergo continued structural rearrangement, and is inherited through unconventional mechanisms during cell division. These properties make ecDNA a highly dynamic form of gene amplification whose effects may extend beyond the genes it carries.

In this thesis, I investigated whether ecDNA does more than simply carry extra copies of cancer-related genes, asking whether its unusual structure, behavior, and maintenance impose specific pressures on cancer cells.

In Chapter 2, we review current knowledge on the generation, maintenance, and elimination of ecDNA. DNA damage, particularly double-strand breaks, plays a central role in ecDNA formation, and multiple DNA repair pathways contribute to the rearrangements that generate ecDNA. Because ecDNA lacks centromeres, it cannot segregate like chromosomes and instead relies on alternative mechanisms for inheritance during cell division. We also discuss its effects on gene expression, its relationship with chromosomal gene amplification, and its frequent localization in micronuclei.

In Chapter 3, we establish controlled model systems in which ecDNA amplifies genes that confer drug resistance rather than oncogenic signaling. These models allow us to distinguish effects associated with ecDNA from those caused by the amplified gene itself. We show that ecDNA is stably maintained under drug selection while remaining structurally dynamic. Transcriptomic analysis reveals only limited shared changes across ecDNA-containing models, suggesting that ecDNA does not induce a strong universal transcriptional program. We also develop a live-cell imaging system that enables fluorescent tracking of ecDNA.

In Chapter 4, we show that ecDNA-positive cells are more sensitive to irradiation-induced DNA damage. This sensitivity is associated with impaired ATM signaling and a weakened G2/M checkpoint, reducing their ability to delay mitotic entry after DNA damage. Radiosensitivity increases with ecDNA abundance, while irradiation is also accompanied by a reduction in ecDNA copy number.

In Chapter 5, comparative genome-wide screens identify selective dependencies of ecDNA-positive cells. Perturbation of ESCRT-0-associated factors, CDAN1, and SIDT2 promotes the accumulation of extranuclear ecDNA-containing structures and selectively reduces the fitness of ecDNA-positive cells. These findings indicate that ecDNA-positive cells depend on pathways involved in controlling the localization, processing, or clearance of aberrantly localized DNA.

Together, these findings show that ecDNA is more than a carrier of amplified genes. Its presence is associated with altered DNA-damage responses and specific cellular dependencies, supporting the idea that ecDNA itself imposes biological constraints on cancer cells and creates vulnerabilities that may be therapeutically exploitable.