Back

Scaling of structural variability of ecDNA polymer condensates with copy number boosts and stabilises oncogene regulatory contacts

Conte, M.; Kundu, S.; Guha, S.; Nicodemi, M.

2026-09-26 biophysics
10.64898/2026.09.25.754334 bioRxiv
Show abstract

Extrachromosomal DNAs (ecDNAs) form highly heterogeneous condensates in cancer cells that drive oncogene overexpression, yet how structural variability coexists with stable gene regulation remains unclear. Here, we develop a minimal polymer physics model of MYC-harbouring COLO320-DM ecDNAs, where BRD4-like complexes bind and bridge cognate sites along ecDNA rings. Above a critical binder concentration, ecDNAs phase separate into condensates exhibiting diverse conformations because of their thermodynamic folding degeneracy. Despite this variability, condensates retain conserved interaction scaffolds that give rise to reproducible contact patterns, including in-trans associated domains (I-TADs), genomic regions enriched in intermolecular regulatory contacts between distinct ecDNAs. We find that condensate 3D architecture follows universal scaling relations with ecDNA copy number, n, remaining robust to model parameter changes. Regulatory contacts within I TADs increase linearly with n, yet they are one order of magnitude stronger than in size matched control regions outside I TADs, whereas their relative fluctuations are markedly suppressed as n increases. This scaling produces enhanced, low-noise regulatory environments for oncogenes embedded within I-TADs, such as PVT1-MYC fusions, whereas the canonical MYC copy, located outside, is less amplified as experimentally observed. Our findings reveal universal polymer physics principles underlying ecDNA condensate organization, offering a mechanistic basis for selective oncogene amplification and potential advantages in cancer progression.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.