BCR-ABL Induces Error-Prone Single Strand Annealing in Transformed Cells.

Chronic myelogenous leukemia (CML) is a hematopoietic stem cell disorder, caused by the BCR-ABL tyrosine kinase oncogene. BCR-ABL kinase activity is required for all aspects of transformation, including abnormal proliferation and neoplastic expansion of stem cells, activation of signaling pathways and genomic instability. We have recently shown that BCR-ABL kinase activity also causes elevated intracellular levels of reactive oxygen species (ROS). These chronically elevated ROS levels have been implicated in the induction of DNA double-strand breaks (DSB) and therapy-related drug resistance, mainly through low- fidelity homology-directed repair (HDR) of DNA lesions. We have utilized GFP-based reporters in BCR-ABL transformed cells to measure both HDR and single-strand annealing (SSA), a mutagenic pathway of homologous repair between repetitive sequences. Repair rates of a single DSB by each of these pathways was measured in BCR-ABL expressing BaF3 cells (BaF3.p210), which have previously been shown to be an excellent system to model the induction of imatinib resistant mutations. We found that inhibition of BCR-ABL kinase activity by imatinib decreased the frequency of SSA by 73 ± 2% (n=3) and increased the frequency of HDR by 70 ± 11% (n=3) in BaF3.p210, compared to untreated cells. Treatment with imatinib did not affect HDR or SSA in BaF3 cells that do not express BCR-ABL. We also determined the fidelity of both SSA and HDR in our model system using clonal populations that stably express the repaired GFP+ reporters (n=20). As expected, sequencing of GFP+ repair products from cells containing the SSA reporter confirmed the expected sequence deletions, consistent with an error-prone mechanism. In contrast, sequencing of GFP+ repair products from the HDR reporter indicated a high-fidelity repair mechanism, without any mutations. It should be noted that this reporter assay does not account for potentially imprecise HDR, leading to a non-functional GFP. Nevertheless, point mutations at the previously reported rate were not detected. Thus, altered HDR fidelity may not be a universal mechanism for the induction of point mutations by BCR-ABL. Our data suggests a unique and alternative pathway, whereby BCR-ABL alters the balance between the SSA and HDR pathways. Moreover, in the presence of interleukin-3 at a concentration that supports not only viability but also cell growth (1ng/ml), imatinib was ineffective at protecting cells from error-prone SSA. In vivo, stromal cells may provide these additional growth signals. Our in vitro data show that stromal cell conditioned medium is not only sufficient to support growth and viability of CML cell lines in the presence of imatinib, but it can also lead to elevated levels of ROS. An abnormal increase in ROS is sufficient by itself to cause genomic mutations (transitions and/or transversions) as a result of single strand oxidative DNA lesions, even in the absence of altered DSB or single strand lesion repair mechanisms. Thus, therapy-related resistance and additional genomic abnormalities may not only be a result of BCR-ABL dependent ROS induction but may also occur within the stromal cell microenvironment through ROS induction by growth factors.