Targeting DNA Repair Gene, RAD52, Induces Exhaustion of the Proliferating CML-CP Leukemia Stem Cells Carrying Oxidative DNA Damage

Abstract 447

BCR-ABL1 transforms hematopoietic stem cells (HSCs) into leukemia stem cells (LSCs) to induce chronic myeloid leukemia in chronic phase (CML-CP). We detected that the most primitive LSCs display elevated levels of reactive oxygen species (ROS) and accumulate excessive numbers of potentially lethal DNA double-strand breaks (DSBs). We also reported that BCR-ABL1-transformed cells exhibit enhanced RAD51-mediated homologous recombination repair (HRR) activity occurring in S and G2/M cell cycle phases. In normal cells initiation of RAD51-mediated HRR is directed either by BRCA1– or RAD52–dependent mechanisms. Since BCR-ABL1 kinase downregulated BRCA1, LSCs containing high number of DSBs should depend more on RAD52 to promote HRR to repair lethal DSBs.

We found that in vivo leukemogenic potential of BCR-ABL1 –positive RAD52−/− hematopoietic cells is abrogated in comparison to their BCR-ABL1 -positive RAD52+/+ counterparts. The absence of RAD52 in BCR-ABL1 –positive cells reduced the percentage of Lin⁻Kit⁺Sca1⁺ cells by >2-fold and inhibited their clonogenic potential and proliferation by >10-fold. In addition RAD52 knockout caused approximately 2-fold reduction of Lin⁻Kit⁺Sca1⁺CD34⁻Flt3⁻ long-term LSCs (LT-LSCs) and Lin⁻Kit⁺Sca1⁺CD34⁺Flt3⁻ short-term LSCs (ST-LSCs). Conversely, 4-fold accumulation of BCR-ABL1 –positive RAD52−/− Lin⁻Kit⁺Sca1⁺eFluor670^max quiescent cells was detected in comparison to BCR-ABL1 –positive RAD52+/+ counterparts. These effects were accompanied by 2-fold reduction of the percentage of BCR-ABL1 –positive RAD52−/− cells in S and G2/M and 7-fold increase of these cells in sub-G1 when compared to BCR-ABL1 –positive RAD52+/+ counterparts. BCR-ABL1-positive RAD52−/− Lin⁻Kit⁺Sca1⁺ cells accumulated more DSBs than BCR-ABL1 –positive RAD52+/+ cells. These differences were not observed between non-transformed RAD52−/− and RAD52+/+ cells. Expression of the wild-type RAD52 reduced the accumulation of lethal DSBs and rescued the clonogenic potential and proliferation of BCR-ABL1-positive RAD52−/− Lin⁻Kit⁺Sca1⁺ cells. Downregulation of ROS with antioxidants vitamin E (VE) and N-acetyl-cysteine (NAC) exerted similar effect as restored expression of RAD52. Thus it appears that RAD52 is necessary to repair the extensive ROS-induced DSBs in LSC-enriched Lin⁻Kit⁺Sca1⁺ cells. BCR-ABL1 kinase does not affect the expression of RAD52 protein, but phosphorylates RAD52 on Y104. However, expression of RAD52(Y104F) phosphorylation-less mutant reduced the number of DSBs and rescued the clonogenic potential of BCR-ABL1-positive RAD52−/− Lin⁻Kit⁺Sca1⁺ cells in a similar way to the wild-type RAD52. Accordingly, RAD52-mediated DSB repair activity in CML-CP cells should not be affected by imatinib treatment. RAD52 mediates the annealing of complementarry DNA strands during DSB repair. To exert this function RAD52 has two DNA binding domains. Expression of RAD52(F79A) and RAD52(K102A) DNA binding-deficient mutants (each amino acid substitution inactivated different DNA binding domain) failed to prevent the accumulation of DSBs and did not rescue the clonogenic and proliferative potential of BCR-ABL1-positive RAD52−/− cells. In addition, RAD52(F79A), but not RAD52(Y104F) inhibited DSB repair by HRR. Therefore DNA binding capability of RAD52 appears essential for BCR-ABL1 –mediated leukemogenesis, but it is dispensable in normal hematopoietic cells. The “addiction” of BCR-ABL1 leukemia cells to RAD52 was confirmed by demonstration that RAD52(F79A) mutant inhibited clonogenic potential of CD34+ CML-CP cells, but not normal counterparts. Furthermore, to determine if RAD52 DNA binding domains could be targeted to selectively inhibit CML-CP, peptide aptamers containing RAD52 DNA binding domain amino acids sequence surrounding F79 were employed as potential decoys for RAD52 DNA binding. Aptamer containing F79, but not the A79 substitution, diminished the number of RAD52 foci and reduced the clonogenic potential and proliferation of CD34⁺ cells from CML-CP, but not from normal donors. In conclusion, we postulate that RAD52 is essential for BCR-ABL1 –mediated leukemogenesis and that DNA binding domains of RAD52 may be targeted for selective elimination of the proliferating CML-CP LSCs.