Oncogene Silencing via ecDNA Micronucleation

Abstract

Extrachromosomal DNA (ecDNA) is a common source of oncogene amplification across many types of cancer. The non-Mendelian inheritance of ecDNA contributes to heterogeneous tumour genomes that rapidly evolve to resist treatment. Here, using single-cell and live-cell imaging, single-micronucleus sequencing, and computational modelling, we demonstrate that elevated levels of ecDNA predisposes cells to micronucleation. Damage on ecDNA, commonly arising from replication stress, detaches ecDNA from the chromosomes upon which they hitchhike during cell division, thereby causing micronucleus formation in daughter cells. Clusters of oncogene-containing, CIP2A-TOPBP1-associated ecDNA molecules form, and asymmetrically segregate into daughter cell micronuclei during cell division. ecDNA chromatin remains highly active during mitosis, but upon micronucleation, it undergoes suppressive chromatin remodeling, largely ceasing oncogene transcription. These studies provide insight into the fate of damaged ecDNA during cell division.

Fig 1.. Cancer cells with ecDNA micronucleate more frequently.

a. Schematic of experimental setup to quantify micronuclei in cancer cell lines and tumours b. Fraction of cells with micronuclei measured in near isogenic cell line pairs (Student’s t-test) c. or primary tumours with or without ecDNA (n = 236, Student’s t-test). d. Fraction of cells with micronuclei in drug-induced ecDNA cell lines treated with either Taxol (T100) or Methotrexate (M600) (One-Way ANOVA). e. Schematic of experimental setup to induce ecDNA in aNSCs and immunoblot of control, induced Myc-ecDNA and control Myc-overexpressing (Myc-OE) aNSCs. f. Fraction of cells with micronuclei control, induced Myc-ecDNA or Myc-OE aNSCs (One-Way ANOVA). g. Exemplary photomicrographs of cells with different ecDNA content and micronuclei (white arrowhead) h-i. Fraction of cells with micronuclei and low (bottom 30%) vs. high (top 30%) ecDNA content measured in (h) cell lines (i) and primary tumours (paired Student’s t-test). j. Fraction of cells with micronuclei combining all cell line and primary tumour data within different ecDNA copy number bins. Percentages of overall copy number were calculated for each dataset and micronucleation rate determined for each copy number bin from lowest (0-5% of overall copy number) to highest (95-100% of overall copy number). k. Exemplary photomicrographs of cells in anaphase with and without detached ecDNA and stained with Hoechst (grey) and MYCN DNA FISH (magenta). l. Number of ecDNA copies per cell with and without detached ecDNA clusters in anaphase (Student’s t-test).

Fig 2.. High ecDNA copy number and targeted DNA damage promote ecDNA micronucleation via detachment in mitosis

a. Schematics to depict replication stress by HU at different time points. b. Quantification of pRPA2 foci per cell in isogenic glioblastoma cell lines at different timepoints of HU treatment in cycling cells (Two-way ANOVA). c. Fractions of cells with micronuclei measured in GBM39ec and GBM39HSR cells (Two-way ANOVA) d. Fraction of total ecDNA in micronuclei (One-way ANOVA) e. Fraction of cells with micronuclei in tumour samples at initial diagnoses and after standard-of care chemotherapy (Student’s t-test). f. Targeted DNA damage inducing by CRISPR on amplicon. Schematics to depict samples collected at different time points. g. Representative images in COLO320DM/COLO320HSR cells to show DNA damage induced at amplicon and amplicon+ micronuclei formation. White arrows indicate micronuclei. h. Total MYC in micronuclei at different time points after nucleofection with CRISPR targeting the amplicon in isogenic cell lines (One way ANOVA) i. Fraction of cells with MYC+ or MYC− micronuclei at different timepoints after nucleofection with CRISPR targeting the amplicon in isogenic cell lines (One-way ANOVA) j. Exemplary photomicrograph of cells in anaphase with detached MYC ecDNA positive for the DNA damage marker γH2AX (Red arrows). Scale bar represents 10 μm k. Untethered fraction of ecDNA in anaphase 24 hrs after nucleofection (Student’s t-test)

Fig. 3:. Clustered ecDNA mis-segregation induces asymmetric mitotic inheritance.

a. Schematics of the experimental setup b. Laser microdissection directed sequencing of primary nuclei (MN) and micronuclei (MN). Counts per million (CPM) – normalized and log2 transformed read coverage (25kb bins) across detected enriched regions and their flanking regions (100 kb). Top annotation: EcDNA reconstruction from long-read sequencing. Regions falling below the detection size limit (75 kb) were excluded (PN: Primary nucleus (n = 5), sMN: single micronucleus (n = 19), pMN: pooled micronuclei (n = 7)) c. Log2-transformed mean coverage ratio of circular reads over the winsorized mean of all linear reads (Welch’s t-test). d. Amplicon-size normalized fraction of circular reads mapping to specific ecDNA species in primary nuclei and micronuclei (Welch’s t-test). e. Amplicon enrichment as determined by FISH in two ecDNA+ cell lines (Student’s t-test). f. Exemplary photomicrographs of a cell with ecDNA in anaphase stained using immunofluorescence against TOPBP1 (yellow) or CIP2A (magenta). g. CIP2a foci increase after 24 hrs HU treatment. h. Exemplary photomicrographs of cells treated for 24 hrs with HU or HU+TD19. i. HU+TD19 treatment increases the number of micronuclei per cell compared to HU only (One-way ANOVA). j. Live-cell imaging of the COLO320DM-Tg19 cell line engineered to visualize MYC ecDNA (TetO-TetR-mNeonGreen) during mitotic transition. Representative snapshots of a mitotic cell (white arrow) carrying an ecDNA-containing micronucleus (yellow arrow) getting asymmetrically inherited into one of the daughter cells upon mitotic exit. Scale bar: 10μm. k. Histograms of ecDNA fractions in the CHP212 cell line in DMSO treated cells without visible lagging cluster (left), DMSO treated cells (middle) and HU treated cells (HU), showing the distribution of ecDNA segregation across daughter cells. 1. Overview and simulation results of the biased random segregation model. m. Density scatter plot of inferred parameters p₁ (extent of biased segregation) and q (proportion of biased segregation). The marginal distributions of p₁ and q are shown above and to the right, respectively. Density values are normalized to 1. n. The inferred proportion of cells exhibiting biased segregation is significantly increased in drug-treated cells compared to untreated cells. Statistical significance was determined using the Mann-Whitney U test (p < 0.001). o. The inferred pi is significantly reduced in drug-treated cells compared to untreated cells. Statistical significance was determined using the Mann-Whitney U test (p < 0.001).

Fig. 4. Reduced oncogenic transcription of ecDNA in micronuclei.

a. Schematic of experimental setup b. Exemplary photomicrographs of cells with ecDNA in anaphase stained using immunofluorescence against H3K27Ac or H3K27me3 (ecDNA is indicated with white arrowheads). c. Ratio of H3K27Ac and H3K27me3 intensity on ecDNA vs. linear chromosomes. d. Exemplary photomicrographs of a cell with ecDNA stained using immunofluorescence against H3K27Ac or H3K27me3 (micronuclei are indicated with arrowheads). e. Hoechst normalized H3K27Ac and H3K27me3 intensity in micronuclei vs. primary nuclei. f. ChIP-sequencing of H3K27ac in COLO320DM cells. Left panel: from top to bottom: peak call annotation. DMSO-subtracted read density as counts per million (CPM). Gene annotation. Right panel: Input-subtracted read density as CPM. g. Exemplary photomicrographs of cells with ecDNA incubated with hydroxyurea (80 μM) or DMSO (vehicle control) and stained using immunofluorescence against RNA polymerase II phosphorylation at serine 5 (micronuclei are indicated with arrowheads). h. Hoechst normalized intensity in micronuclei vs. primary nuclei from cells stained using immunofluorescence against RNA polymerase II phosphorylation at serine 5. i. Exemplary photomicrographs of COLO320DM cells and their micronuclei (white dashed line) labelled with intron MYC RNA FISH signal (left), showing a non-transcribing micronucleus (top) and an actively transcribing micronucleus (bottom). A quantification of the number of transcribing and non-transcribing micronuclei in COLO320DM (n=102), GBM39EC (n=109), and PC3DM (n=104) (right). Scale bar: 10μm. j. Z-score normalized transcript counts of the circularly amplified MYC oncogene in COLO320DM cells. Three replicates per treatment condition. Benjamini-Hochberg procedure-corrected p-value is shown (Wald test). k. Gene set enrichment analysis (GSEA) of MYC target genes. Benjamini-Hochberg procedure-corrected p-value is shown. l. Schematic highlighting the main findings of this publication