Several mutations are required for cancer development, and genome sequencing has revealed that many cancers, including breast cancer, have somatic mutation spectra dominated by C-to-T transitions. Most of these mutations occur at hydrolytically disfavoured non-methylated cytosines throughout the genome, and are sometimes clustered. Here we show that the DNA cytosine deaminase APOBEC3B is a probable source of these mutations. APOBEC3B messenger RNA is upregulated in most primary breast tumours and breast cancer cell lines. Tumours that express high levels of APOBEC3B have twice as many mutations as those that express low levels and are more likely to have mutations in TP53. Endogenous APOBEC3B protein is predominantly nuclear and the only detectable source of DNA C-to-U editing activity in breast cancer cell-line extracts. Knockdown experiments show that endogenous APOBEC3B correlates with increased levels of genomic uracil, increased mutation frequencies, and C-to-T transitions. Furthermore, induced APOBEC3B overexpression causes cell cycle deviations, cell death, DNA fragmentation, γ-H2AX accumulation and C-to-T mutations. Our data suggest a model in which APOBEC3B-catalysed deamination provides a chronic source of DNA damage in breast cancers that could select TP53 inactivation and explain how some tumours evolve rapidly and manifest heterogeneity.
Bibliographical noteFunding Information:
Acknowledgements We thank J. Hultquist and R. Vogel for statistics, T. Hwang for bioinformatic assistance, V. Polunovsky for hTERT-HMECs, V. Simon for shRNA, S. Kaufmann, C. Lange and D. Largaespada for consultation, and the Masonic Cancer Center Breast Cancer Research Fund for purchasing the ATCC breast cancer panel. Tissues were obtained from the Masonic Cancer Center Tissue Procurement Facility, which is part of BioNet, supported by the Academic Health Center and National Institutes of Health (NIH) grants P30 CA77598 (D.Y.), P50 CA101955 (D. Buchsbaum) and KL2 RR033182 (B. Blazar). M.B.B. was supported in part by a Cancer Biology Training Grant (NIH NCI T32 CA009138) and a Department of Defense Breast Cancer ResearchProgram Predoctoral Fellowship (BC101124). L.L. was supported in part by a National Science Foundation Predoctoral Fellowship and by a position on the Institute for Molecular Virology Training Grant NIH T32 AI083196. M.A.C. was supported by an NIH postdoctoral fellowship (F32 GM095219). A.M.L. was supported by a CIHR postdoctoral fellowship. E.W.R. was supported by a position on the Institute for Molecular Virology Training Grant NIH T32 AI083196 and subsequently by an NIH predoctoral fellowship (F31 DA033186). Computational analyses (N.A.T. and D.E.D.) were supported by federal funds from the National Cancer Institute, NIH, CBIIT/caBIG ISRCE yellow task 09-260. The Harris laboratory was supported in part by NIH R01 AI064046, NIH P01 GM091743, the Children’s Cancer Research Fund, and a seed grant from the University of Minnesota Clinical and Translational Science Institute (supported by NIH 1UL1RR033183).