Evolvability of cancer-associated genes under APOBEC3A/B selection

Abstract

Evolvability is an emergent hallmark of cancer that depends on intra-tumor heterogeneity and, ultimately, genetic variation. Mutations generated by APOBEC3 cytidine deaminases can contribute to genetic variation and the consequences of APOBEC activation differ depending on the stage of cancer, with the most significant impact observed during the early stages. However, how APOBEC activity shapes evolutionary patterns of genes in the host genome and differential impacts on cancer-associated and non-cancer genes remain unclear. Analyzing over 40,000 human protein-coding transcripts, we identified distinct distribution patterns of APOBEC3A/B TC motifs between cancer-related genes and controls, suggesting unique associations with cancer. Studying a bat species with many more APOBEC3 genes, we found diverse motif patterns in orthologs of cancer genes compared to controls, similar to humans and suggesting APOBEC evolution to reduce impacts on the genome rather than the converse. Simulations confirmed that APOBEC-induced heterogeneity enhances cancer evolution, shaping clonal dynamics through bimodal introduction of mutations in certain classes of genes. Our results suggest that a major consequence of the bimodal distribution of APOBEC affects greater cancer heterogeneity.

Keywords: APOBEC; Evolvability; Intra-tumor heterogeneity; Neutral evolution.

Figure 1.

APOBEC3A/B TC hotspots statistics of protein-coding transcripts of the human genome. a) CDUR analysis on human genome shows two dense regions at top-left and right-bottom extremes and relatively sparse dispersion elsewhere. b) Genes in the top-left partition of the CDUR plot showed significant enrichment of GO Biological Processes associated with differentiation and development. c) Genes in the top-left partition of the CDUR plot showed significant enrichment of KEGG Pathway associated with several cancers, basal cell carcinoma, breast cancer and gastric cancer.

Figure 2.

Cancer-associated genes have significantly different distribution in CDUR plot compared to control genes. a) Cancer-associated genes and control genes show distinct distribution in both motif under-representation and mutational susceptibility (Kolmogorov-Smirnov test for motif-representation axis, P = 2.828 × 10⁻⁷; for mutational susceptibility axis, P = 1.175 × 10⁻⁴). b) Motif under-representation and mutational susceptibility of BRCA1 orthologs show higher variance than sequential mutations of the gene on CDUR plot. c) Standard deviation of motif under-representation comparison between orthologs and sequential mutations show significantly high variance among orthologs in both cancer-associated genes (196 genes) and control genes (20 genes). d) Standard deviation of mutational susceptibility comparison between orthologs and sequential mutations show significantly high variance among orthologs in both cancer-associated genes (196 genes) and control genes (20 genes).

Figure 3.

APOBEC3A/B TC hotspots statistics of protein-coding transcripts of Pteropus alecto genome. a) CDUR analysis on bat genome shows two dense regions at top-left and right-bottom extremes and relatively sparse dispersion elsewhere as in the human genome. b) Bat orthologs of human cancer-associated genes and control genes show distinct distribution in mutational susceptibility but not in motif under-representation and (Kolmogorov-Smirnov test for motif-representation axis, P = 0.1701; for mutational susceptibility axis, P = 3.169 × 10⁻⁵)

Figure 4.

Protein-coding transcripts from human genome display biased distribution in APOBEC3A/B motif representation and mutational susceptibility. Bimodal distribution of APOBEC3A/B motif representation displays higher heterogeneity by spatial clonal dynamics simulation. a) The number of genotypes between with and without APOBEC3A/B shows no significant difference. b) Heterogeneity between the two simulations with and without APOBEC3A/B shows no significant difference. c, d) Muller plots of simulation with and without APOBEC3A/B activity shows no significant difference in heterogeneity at the final time point. e) The number of genotypes between uniform and bimodal distribution shows an increasing difference over time. f) Heterogeneity between uniform and bimodal distribution shows an increasing difference over time. g, h) Muller plots of simulation uniform and bimodal distribution show an increasing difference over time.