Phase separation of ecDNA aggregates establishes in-trans contact domains boosting selective MYC regulatory interactions

Abstract

Extrachromosomal DNAs (ecDNAs) are found in the nucleus of an array of human cancer cells where they can form clusters that were associated to oncogene overexpression, as they carry genes and cis-regulatory elements. Yet, the mechanisms of aggregation and gene amplification beyond copy-number effects remain mostly unclear. Here, we investigate, at the single molecule level, MYC-harboring ecDNAs of COLO320-DM colorectal cancer cells by use of a minimal polymer model of the interactions of ecDNA BRD4 binding sites and BRD4 molecules. We find that BRD4 induces ecDNAs phase separation, resulting in the self-assembly of clusters whose predicted structure is validated against HiChIP data (Hung et al., 2021). Clusters establish in-trans associated contact domains (I-TADs) enriched, beyond copy number, in regulatory contacts among specific ecDNA regions, encompassing its PVT1-MYC fusions but not its other canonical MYC copy. That explains why the fusions originate most of ecDNA MYC transcripts (Hung et al., 2021), and shows that ecDNA clustering per se is important but not sufficient to amplify oncogene expression beyond copy-number, reconciling opposite views on the role of clusters (Hung et al., 2021; Zhu et al., 2021; Purshouse et al. 2022). Regulatory contacts become strongly enriched as soon as half a dozen ecDNAs aggregate, then saturate because of steric hindrance, highlighting that even cells with few ecDNAs can experience pathogenic MYC upregulations. To help drug design and therapeutic applications, with the model we dissect the effects of JQ1, a BET inhibitor. We find that JQ1 reverses ecDNA phase separation hence abolishing I-TADs and extra regulatory contacts, explaining how in COLO320-DM cells it reduces MYC transcription (Hung et al., 2021).

Figure 1.. Ring polymer model of ecDNA of COLO320-DM colorectal cancer cells.

A) The genomic arrangement of chromosomal fragments of the MYC-harboring ecDNA of COLO320-DM colorectal cancer cells and of its BRD4 binding sites (magenta circles) is mapped at 50kb resolution (Hung et al., 2021). B) Our Strings and Binders (SBS) polymer model of the ecDNA consists of a ring of beads with BRD4 binding sites (magenta) and BRD4 molecules (green) that can bridge them. Here a linear map is shown of BRD4 sites of the polymer model of the ecDNA, magenta shades represent different affinity strengths. Below are mapped the copies of MYC (MYC-1, −2 and −3, red), PVT1 (PVT1–1, −2, −3, and −4, blue) and cis-regulatory elements (yellow) (Hung et al., 2021). C) A cartoon of the polymer model and how it folds by interaction with BRD4 molecules.

Figure 2.. Single rings have a coil-globule transition driven by interactions with BRD4.

A) In a model including only one ring, above a threshold concentration of BRD4, the polymer has a phase transition from a coil to a globular state. B) That is shown by a sharp drop of its average gyration radius. C,D) This structural change also occurs in a system with multiple rings. E, F) The average distance and contact matrices of a one ring model are shown in its coil (no BRD4, top) and globular state ([BRD4]=50nmol/l, bottom). G) A single-molecule ring 3D structure in the two phases.

Figure 3.. Phase separation of ecDNAs establishes in-trans associated contact domains.

A) A system of multiple rings phase separates into a single cluster when BRD4 grows above a threshold concentration. B,C) That is marked by a sharp drop of the average distance of their centers of mass, in a system with resp. 2 and 10 rings. D) The average distance and E) the in-cis, F) in-trans and G) total contact matrices per polymer are shown in a system of 2 and 10 rings (resp. top and bottom) in the phase separated state. In a cluster of rings promiscuous contacts are not random, as specific in-trans associated contact domains (I-TADs, black dashed squares in the figure) appear between strong BRD4 binding sites. MYC-2/−3 and PVT1-2/−3/−4 are part of I-TADs and enriched of regulatory contacts, while MYC-1 and PVT1-1 are not, despite belonging to the cluster too. H) A single-molecule 3D structure of a cluster of 2 and 10 rings in the model phase separated state. The average diameter of a 10 rings cluster in 500nmol/l of BRD4 is around 850nm. The number of contacts plateaus in clusters larger than half a dozen rings because of steric hindrance.

Figure 4.. The model predicted 3D structure of clusters is validated by HiChIP data.

A) H3K27ac HiChIP bulk data are shown for the ecDNA in COLO320-DM colorectal cancer cells (Hung et al., 2021). B) The total contact matrix per ring of phase separated clusters of 10 polymers (bottom Fig. 3G) is shown here remapped on the WT genome. The Pearson and genomic-distance corrected Pearson correlations of the two matrices are r=0.7 and r’=0.6, supporting the view that the model basic ingredients do capture important elements in the self-assembly of ecDNA clusters.

Figure 5.. ecDNA clusters selectively boost MYC-2/−3 promiscuous regulatory contacts, not MYC-1.

A) As the number of rings in a phase separated cluster increases, the virtual 4C contact profile (per ring) from the viewpoint of MYC-1 only marginally grows above the baseline (black dashed line, contacts formed by MYC-1 in a 1 ring model with no BRD4). B) Conversely, the promiscuous contacts of MYC-2 markedly increase as it becomes part of I-TADs, because it is close to a strong BRD4 site (analogously MYC-3, not shown). That illustrates that cluster phase separation has different effects on the contacts of different gene copies on the ecDNA, depending on the underlying genomic arrangement and affinity of BRD4 sites. C) The change of regulatory contacts relative to regions devoid of BRD4 sites is shown as the number of rings in a cluster varies. MYC-2 and PVT1–2 have a threefold increase of contacts as the number of rings grows from 1 to 10. The change plateaus above half a dozen rings because of steric hindrance effects. MYC-1 and PVT1–1, instead, are not different from the control, showing that ecDNA clusters selectively boosts beyond copy number effects MYC-2 and MYC-3, but not MYC-1 regulatory interactions. D) In a model system with 5 rings, the fraction of MYC regulatory contacts associated to PVT1-MYC fusions (i.e., MYC-2 and MYC-3) is 68% of the total (including also two extra canonical copies of MYC to account for chromosomal genes). E) In COLO320-DM cells, where the mean number of ecDNAs in a cluster is five, PVT1-MYC fusions account for 73% of MYC transcripts (Hung et al. 2021). Those results combined hint that the oncogene transcription boost is controlled by the increase of its regulatory contacts.

Figure 6.. JQ1 reverses phase separation of ecDNAs and erases their contact domains.

A) In our polymer model, the BET inhibitor JQ1 is represented as a new type of molecule that strongly antagonizes ecDNA BRD4 binding sites for BRD4, hence preventing the latter from bridging ecDNAs. The effects of JQ1 are illustrated in a system with 10 rings and [BRD4]=500nmol/l. As JQ1 concentration increases, the virtual 4C contact profile (per ring) from the MYC-1 and, B), MYC-2 viewpoint collapse to the baseline. C) JQ1 has a non-linear impact on the relative variation of gene regulatory contacts (per gene copy): by reducing the concentration of free BRD4 below threshold it reverses phase separation, hence dissolving ecDNA clusters and their I-TADs. D) In a model with five rings, when the concentration of JQ1 goes from zero to 500nmole/l, MYC total regulatory contacts (including two additional canonical copies of MYC accounting for chromosomal genes) have a 69% drop. E) In COLO320-DM cells when JQ1 grows from zero to 500nmole/l, MYC transcription probability falls by around 71% (Hung et al. 2021). Our results combined hint that MYC overexpression, beyond copy number effects, is mainly driven by regulatory contacts enhanced by ecDNA clusters.