Overall, we propose that ecDNA is not only selected in cancer because of oncogenes, it is also actively retained because of retention elements, allowing it to persist in a growing cancer cell population. 10/11

Finally, retention elements are unmethylated on ecDNA and when we methylate them, ecDNA becomes untethered from chromosomes. Plasmids with methylated retention elements are also no longer retained. 9/11

Using live cell imaging, we found that adding a retention element to a plasmid makes it more likely to attach to chromosomes during mitosis. 6/11

We verified that these elements, when individually cloned into a plasmid, can promote retention of the plasmid in cells. Putting more copies of an element in a plasmid increases its retention additively. 5/11

To identify elements within ecDNA that may act as anchors for hitchhiking, we shattered the entire human genome into short pieces of DNA, cloned them into a plasmid pool and asked which DNA pieces can be retained over many cell passages. 3/11

ecDNA does not have centromeres but is inherited along with chromosomes by tethering to them. Although this “hitchhiking” phenomenon has been observed for many years, the mechanism is unknown. 2/11

Honored to receive the Science & SciLifeLab Prize! Very grateful to all my mentors and colleagues. I wrote an article in @science.org about my PhD thesis research on extrachromosomal DNA at Stanford @stanfordmedicine.bsky.social. www.science.org/doi/10.1126/...

Cool approach to isolate extrachromosomal DNA: karyotyping by FACS! Wish I had thought of it. www.biorxiv.org/content/10.1...

Overall, these results show unique distribution and evolution of extrachromosomal oncogene copies in cancer cells. Co-inheritance allows cancer cells to keep specific combinations of DNA sequences, allowing the winning combination to get amplified in cancer. 8/

Co-inheritance leads to coordinated dynamics of ecDNAs in cells under selective pressure. Treating cancer cells with a drug that inhibits one of the oncogene products leads to co-depletion of multiple ecDNAs, suggesting that they can have coupled responses to perturbations. 7/

While evolutionary modeling suggested that cells harboring multiple ecDNA species could have been selected, the observed copy number correlation is mainly explained by co-inheritance rather than co-selection of ecDNAs. 6/

We think co-inheritance of ecDNAs is promoted by transcription complexes. Inhibition of transcription initiation (but not elongation) by triptolide reduced co-inheritance. 5/

Surprisingly, the answer was no. Rather, we found that different ecDNAs containing cancer-driving genes are passed down together when cancer cells divide, leading to correlations in combinations of ecDNAs. 4/

Normally, DNA is stably inherited when cells divide. Cancer cells bend this rule by keeping oncogenes outside chromosomes on ecDNA. As cancer cells can have many copies and distinct sequences of ecDNAs, we asked whether these ecDNA species are randomly distributed among cells. 3/