Evolution of oncogene amplification across 86,000 cancer cell genomes

Abstract

High-level copy-number (CN) amplification (HLAMP) is a major mechanism of oncogene activation in human cancer. Despite progress in therapeutically targeting HLAMPs, the processes underlying HLAMP evolution are incompletely understood, leaving critical knowledge gaps in their etiology and mechanisms of therapeutic response. To address this, we analyzed using novel computational approaches, the evolutionary trajectories of HLAMP in single-cell whole-genome sequencing of 86,239 cancer cells from 94 patients and 8 experimental systems. We found that two common etiologies of oncogene amplification-intrachromosomal amplification (ICamp) and extrachromosomal circular DNA (ecDNA)-lead to distinct CN distributions across cells in amplitude, variance and clone specificity. Notably, ICamp events exhibited widespread subclonal specificity, indicating their dynamic evolution through numeric or structural modulation associated with clonal expansions. In contrast, ecDNAs exhibited higher cell-to-cell CN variation linked to structural rearrangements, whereby complex ecDNA architectures were resolved at single-nucleotide resolution in individual cell genomes. Through joint analysis of ecDNA structure and their cellular abundance, we observed both divergent and convergent modes of ecDNA evolution shaped by DNA damage, selection, and chromosomal re-integration. Finally, modeling single-cell distributions of ecDNAs substantially improved bulk genome-graph based predictions, revealing that ecDNA prevalence is tissue-type specific, rather than broadly shared across cancer types, with implications for ecDNA-directed therapy.

Fig. 1.. Cell-to-cell copy-number variance in scWGS differentiates ecDNA-mediated amplifications from intrachromosomal amplifications.

(A) Cancer types and case counts (top) and schematic illustration of study design (bottom). Organ and model illustrations are from BioRender. (B) Number of high-quality cancer cells analyzed (x-axis, log-scale) by tumor type (y-axis). (C) Frequently amplified oncogenes and their case counts. (D) Inheritance patterns for intrachromosomal amplification (ICamp) versus extrachromosomal circular DNA (ecDNA; left) and expected CN distributions across cells (right). (E) Representative examples of ICamp and ecDNA, from a patient with EGFR-mutated lung adenocarcinoma (left) and glioblastoma (right). (F) Conceptual illustration of two metrics used to classify HLAMP events. (G) Classification of 507 HLAMP regions from 86,239 cancer cells based on single-cell CN distribution. (H) Fraction of clonal and subclonal HLAMP regions by mechanism, across cases.

Fig. 2.. Contrasting mechanisms of ICamp CN heterogeneity by numeric and structural modulators.

(A) Prevalence of multi-peak CN distribution among ICamp cases. (B) An ICamp case exhibiting multi-peak CN distribution across cells, indicating subclones with distinct ERBB2 CN. (C) CN variance of HLAMP events stratified by initiating mechanism (BFB, breakage-fusion bridge; CGR-NOS, complex genomic rearrangement, not otherwise specified; CGR-TRA, complex genomic rearrangement, translocation-rich). (D) Initiating and modulating mechanisms for ICamps. (E) Pairwise comparisons of HLAMP-associated arm-level CN profiles between subclones. (F) A classical example of aneuploidy-mediated doubling of CCNE1 amplification in EL011. Compared to clone C, all the HLAMP segments exhibit proportional CN increase in clone H. (G) A structural modulation of 17q amplification in OV-002. Clone B exhibits further amplification of two segments (one amplifying MIR21) within the HLAMP segment in clone A. (H) Multiple CN peaks of CCND1 amplification in OV-052 matching the subclonal phylogeny (left), clone-specific CNs and SVs indicative of BFB-mediated amplification and subsequent modulation by arm-level aneuploidy (middle), and schematics of the mechanism (right). (I) An ICamp event involving chromosome 8 in OV-083, initiated by chromothripsis, followed by a second chromothripsis and subsequent whole-chromosome aneuploidy.

Fig. 3.. Resolving mechanisms of segregation of co-amplified oncogenes in ecDNA in single cells.

(A, B) Genome-wide CN profiles (left) and CN correlations between key oncogenes (right) for GBM0721 (A) and EL001 (B). (C, D) Single cell CN/SV profiles indicating two distinct simple ecDNAs in GBM0721 (C) and two distinct hybrid ecDNAs in EL001 (D). (E) Hierarchical clustering of HLAMP regions in EL001 based on pairwise CN correlations reveals two ecDNA species co-amplifying oncogenes. (F) Decomposition of ecDNA heteroplasmy in EL001 by ECADeMix. HLAMP segments used (S1-S4) are indicated in D. (G) Pseudobulk CN/SV profile from Lx516 showing multi-chromosomal chromothripsis leading to HLAMP of KDM2A. (H) Clustering based on pairwise CN correlation identifies a hybrid ecDNA and a subclonal ICamp event. Rearrangements between segments are illustrated as arcs on the top. Chromosomal origins in the left heatmap. Subclonal CN and ecDNA scores on the right side. Inset plots show the reconstructed ecDNA structure (left) and a representative CN correlation between two segments that comprise the ecDNA. (I) Schematic illustration of HLAMP mechanisms in Lx516.

Fig. 4.. Modes of ongoing ecDNA evolution.

(A) Chromothripsis-derived ecDNA amplifying MYCN in the NCI-H69 SCLC cell line (top left); single-cell CN profiles of the amplicon and ecDNA heteroplasmy delineated by ECADeMix (bottom left). CN/SV profiles from 4 cancer cells with distinct ecDNA subspecies (right). e0 is the likely parental ecDNA; e1-e3 are daughter ecDNAs derived from e0 via internal rearrangements. (B) Schematic illustration of branched evolution via ecDNA rearrangement. (C, D) Pseudobulk genome-wide CN profile of glioblastoma GBM0510, exhibiting multiple ecDNA-mediated HLAMPs (C, top left); single-cell CN profiles of the amplicons and decomposed ecDNA species (C, bottom left). EGFR is amplified by three distinct ecDNAs (e1, e2, and e3) in a mutually exclusive manner across cells, forming distinct subclones (single-cell examples in the right panel of C). ecDNAs amplifying TERT and CDK4 are monotypic, likely reflecting temporal order among the events (D).

Fig. 5.. scWGS distinguishes current ecDNA from historical bulk-level footprints.

(A) Case-level comparison of ecDNA predictions from bulk genome-graph methods (Amplicon Architect/Classifier) versus scWGS. (B) HLAMP region-level comparison divided by cancer type. (C) Junctional CN difference profiles in ecDNA and ICamp regions. Statistical significance was determined by Student’s t test. (D) CN distributions in isogenic paired cell lines: ecDNA-mediated double minutes (COLO320-DM and GBM39-DM) versus homogeneously staining regions (COLO320-HSR and GBM39-HSR). (E) Cytogenetic features of GBM39 isogenic pair exhibiting numerous variable sized, dispersed signals consistent with EGFR-bearing ecDNA (top) and intrachromosomal clustered amplification manifesting as homogeneously staining region (HSR). (F, G) Pseudobulk CN/SV profiles (top) and single-cell CN profiles (bottom) for DM versus HSR lines reveal genomic rearrangements that occurred after the most recent common ancestor and prior to HSR formation in COLO320 (F) and GBM39 (G) isogenic pairs.