RAD52-Dependent Synthetic Lethality Eradicates Leukemia Stem Cells

Abstract 3

We showed that imatinib-naive and imatinib-treated BCR-ABL1 kinase-positive chronic myeloid leukemia in chronic phase (CML-CP) Lin⁻CD34⁺CD38⁻ stem cells (LSCs) and Lin⁻CD34⁺CD38⁺ leukemia progenitor cells (LPCs) accumulate high numbers of the reactive oxygen species (ROS)-induced DNA double-strand breaks (DSBs) which, if not repaired, are lethal (Nieborowska-Skorska et al., Blood, 2012). Genome-wide microarray screen indicated that LSCs overexpress numerous genes responsible for homologous recombination repair (HRR) of DSBs. Direct targeting of RAD51, a key element in HRR, by small molecule inhibitor exerted anti-CML effect, but normal cells were also affected, indicating that RAD51 is not a preferable target. In normal cells HRR depends primarily on BRCA1/BRCA2-RAD51 pathway, while RAD52-RAD51 pathway may serve as a back-up. However BRCA1 protein is downregulated in CML, and RAD51(F259V) mutant (RAD52 binding-deficient) induced apoptosis and reduced cell growth in BCR-ABL1 –dependent manner, thus highlighting the potential role of alternative RAD52-driven HRR in CML. The absence of RAD52 (Rad52−/−) inhibited BCR-ABL1 –mediated cell cycle progression and clonogenic activity, induced apoptosis, reduced Lin⁻Kit⁺Sca1⁺CD34⁻Flt3⁻ long-term LSCs and Lin⁻Kit⁺Sca1⁺CD34⁺Flt3⁻ short-term LSCs, and abrogated leukemogenesis in murine model of CML. At the same time RAD52 was expendable in normal cells. In addition, RAD52 was essential to prevent accumulation of ROS-induced lethal DSBs in BCR-ABL1 –positive murine LSCs. Thus it appears that RAD52 is necessary to repair numerous ROS-induced DSBs in LSCs to promote leukemogenesis. BCR-ABL1 kinase interacts with and phosphorylates RAD52 on Y104, but phosphorylation-deficient RAD52(Y104F) mutant did not exert any negative effect on BCR-ABL1 kinase-driven leukemogenesis. This observation suggests that RAD52 activity is preserved in imatinib-treated cells. On the other hand, DNA binding-deficient RAD52(F79A) and RAD52(K102A) mutants (disrupt DNA binding domain I and II, respectively) inhibited clonogenic potential of Lin⁻CD34⁺ CML-CP cells and BCR-ABL1 –positive Lin⁻Kit⁺Sca1⁺ murine LSCs, but not normal counterparts. Therefore, RAD52-mediated DNA binding activity appears to be essential for BCR-ABL1 kinase-driven leukemogenesis, but not normal hematopoiesis. RAD52 DNA I binding groove containing F79 forms a pocket/niche. Peptide aptamer containing amino acid residues surrounding RAD52 F79, but not that with F79A substitution, inhibited RAD52 foci, RAD52-dependent RAD51 foci and HRR activity, resulting in elevation of the number of lethal DSBs and abrogation of clonogenic activity of BCR-ABL1-positive murine leukemia cells. Aptamer F79 reduced the growth of LSCs and LPCs from CML-CP and more aggressive CML-accelerated phase (CML-AP), depleted quiescent LSCs by attrition and eradicated BCR-ABL1 leukemia from SCID mice. Moreover, aptamer F79 enhanced the anti-CML effects of imatinib. The effect of aptamer F79 in BCR-ABL1 leukemia cells depended on downregulation of BRCA1 implicating “synthetic lethality”. At the same time the aptamer did not exert any measurable negative effects on normal cells and tissues. In conclusion, we postulate that CML-CP/AP cells are “addicted” to RAD52 and that targeting of DNA binding domain of RAD52 may induce “synthetic lethality” to eliminate proliferating LSCs/LPCs and to deplete quiescent LSCs. Moreover, anti-RAD52 F79 aptamer exerted anti-tumor activity in acute myeloid leukemia primary cells, and in cell lines derived from carcinomas of breast, ovary and pancreas displaying BRCA1 and/or BRCA2 deficiency. Thus, BRCA1/BRCA2ness-driven “addiction” to RAD52 may be selectively targeted to achieve “synthetic lethality” in wide-range of tumors while being harmless to normal cells.