BCL2 suppresses PARP1 function and nonapoptotic cell death

Abstract

BCL2 suppresses apoptosis by binding the BH3 domain of proapoptotic factors and thereby regulating outer mitochondrial membrane permeabilization. Many tumor types, including B-cell lymphomas and chronic lymphocytic leukemia, are dependent on BCL2 for survival but become resistant to apoptosis after treatment. Here, we identified a direct interaction between the antiapoptotic protein BCL2 and the enzyme PARP1, which suppresses PARP1 enzymatic activity and inhibits PARP1-dependent DNA repair in diffuse large B-cell lymphoma cells. The BH3 mimetic ABT-737 displaced PARP1 from BCL2 in a dose-dependent manner, reestablishing PARP1 activity and DNA repair and promoting nonapoptotic cell death. This form of cell death was unaffected by resistance to single-agent ABT-737 that results from upregulation of antiapoptotic BCL2 family members. On the basis of the ability of BCL2 to suppress PARP1 function, we hypothesized that ectopic BCL2 expression would kill PARP inhibitor-sensitive cells. Strikingly, BCL2 expression reduced the survival of PARP inhibitor-sensitive breast cancer and lung cancer cells by 90% to 100%, and these effects were reversed by ABT-737. Taken together, our findings show that a novel interaction between BCL2 and PARP1 blocks PARP1 enzymatic activity and suppresses PARP1-dependent repair. Targeted disruption of the BCL2-PARP1 interaction therefore may represent a potential therapeutic approach for BCL2-expressing tumors resistant to apoptosis.

Figure 1. Nuclear BCL2 interacts with PARP1

A. Nuclear (N) and cytoplasmic fractions from Eμ-MYC/MMTV-BCL2 B-cell acute lymphoblastic leukemias (1863 and 3256) (15) were immunoblotted for BCL2, tubulin (cytoplasmic marker) and Histone H3 (nuclear marker). DLBCL lines were analyzed by immunoblotting 1 hour after irradiation or treatment with MNNG in the presence or absence of ABT-737. B. HEK293T cells were transfected with an expression plasmid for HA-BCL2 and fractionated lysates were collected after 24 hours. C. OCI-LY8 cells were exposed to ionizing radiation. After 30 minutes, whole cell lysate (WCL) or subcellular fractions were isolated and subjected to immunoprecipitation with anti-BCL2 antibody followed by separation on SDS-PAGE gel and Coomassie staining. Mass spectrometry of the ∼113 kDa band (top arrow) revealed multiple peptides from PARP1. Mass spectrometry of the lower two bands indicated by arrows identified multiple histones. D. OCI-LY8 cells were irradiated and then treated with vehicle (DMSO), ABT-737 (737), an ABT-737 enantiomer (enant), or ABT-888 (888). Subcellular fractions were immunoprecipitated with anti-BCL2 antibody and immunoblotted for BCL2, PARP1 or KU70. Loading and fractionation were confirmed by immunoblotting (Input). WCL indicates whole cell lysate. E. Immunoblotting against PAR in the presence or absence of 100 μM MNNG, 100 nM 100 nM ABT-888, the combination or 100 nM ABT-737.

Figure 2. ABT-737 disrupts the BCL2-PARP1 interaction

A. Purified PARP1 was incubated with GST-tagged BCL2, GST-BCL-xL or GST–BCL-w. Mixtures were subjected to immunoprecipitation with anti-GST antibody followed by immunoblotting for PARP1. Ponceau staining was performed on immunoprecipitates to confirm equal recovery. B. Fluorescence polarization assay of GST-BCL2 mixed with increasing concentrations of ABT-737, NOXA or PARP1, as previously described (22). Error bars indicate standard error of measurement. C. GST-BCL2 and PARP1 were incubated along with increasing concentrations of ABT-737. Immunoprecipitation and immunoblotting were performed, as in (A). D. Schematic representation of full-length PARP1, deletion constructs and brct domain constructs. F1 and F2 are zinc finger 1 and zinc finger 2 domains, respectively. N indicates the nuclear localization signal. E. Bacterially expressed PARP1 proteins were separated on SDS-PAGE gel and immunoblotted with anti-PARP1 antibody. Arrows indicate the expected sizes. F. His-PARP1 contructs were incubated with purified GST-BCL2 followed by immunoprecipitation with anti-His antibody and immunoblotting against BCL2. Purified GST-BCL2 was used as a control for BCL2 immunoblotting. Incubation of the His-ΔBRCT and GST-BCL2 was performed in the presence or absence of 100 nM ABT-737. G. GST-BCL2 was incubated with brct-only or brctΔBH3 constructs followed by immunoprecipitation and immunoblotting with an antibody specific for the PARP1 BRCT domain. Ponceau staining was performed to confirm equal loading.

Figure 3. BCL2 suppresses PARP1 enzymatic function

A. Purified GST-BCL2 was added to fractionated lysates from HT cells or ABT-737 was added to fractionated lysates from OCI-LY8 cells. PARP1 activity on immobilized histones was measured by enzyme linked immunosorbent assay. Error bars represent standard error of measurement. B. GST-MCL1, GST-BCL-xL and GST-BCL-w were added to nucleoplasm extracts from HT cells and PARP1 activity was assayed, as above. C. OCI-LY1-10R cells treated with MNNG for 15 minutes were washed and exposed to vehicle (DMSO), ABT-888 or ABT-737. Cells were collected prior to MNNG treatment (pre), immediately after washing MNNG (post MNNG) or 3 hours after wash. Olive moments were quantified for 500 cells in each category. Error bars indicate standard error of measurement. Pictures are 100X magnification.

Figure 4. BCL2 suppresses MNNG-induced death in DLBCL cells

A. Cells were treated with indicated agents for 20 hours and analyzed by flow cytometry after staining with propidium iodide (PI) and FITC-conjugated annexin V. Percent alive is the fraction of Annexin V/PI-negative cells relative to the same cell line treated with DMSO and no MNNG. Error bars indicate standard deviation. * indicates p<0.05. B. Example of flow cytometric plots of OCI-LY1-10R cells treated with agents for 20 hours and stained with PI and FITC-annexin V. C. Transmission electron micrographs of OCI-LY1-10R cells fixed 9 hours after treatment with DMSO, 500 μM MNNG, 100 nM ABT-737, 50 μM etoposide or combinations.

Figure 5. ABT-737 promotes MNNG-induced cell death in BCL2-overexpressing cells that is independent of BAX/BAK

A. Bax^-/-Bak^-/- MEFs (DKO) and BCL2-transgenic Bax^-/-Bak^-/- MEFs (BCL2-DKO) were treated with indicated agents for 20 hours and analyzed by flow cytometry after staining with propidium iodide (PI) and FITC-conjugated annexin V. Percent alive is the fraction of Annexin V/PI-negative cells relative to the same cell line treated with DMSO and no MNNG. Error bars indicate standard deviation. * indicates p<0.05. B. Transmission electron micrographs at 890X and 2900X magnifications fixed 9 hours after treatment with the indicated agents. C. Concentrations of NAD+ and ATP were determined in lysates isolated after 9 hour treatment with MNNG and the indicated agents. All values are relative to the same cell line treated with vehicle and no MNNG.

Figure 6. ABT-737 promotes non-apoptotic death in primary CLL cells independent of stroma

A. CLL cells were cultured with or without primary marrow stromal cells (HMSCs) and surface CXCR4 expression was quantified by flow cytometry. B. Mitochondria from six CLL samples were exposed to ABT-737 and the relative depolarization was quantified based on cytochrome C release, as previously described (32). FCCP was used as a positive control. C. CLL cells from patient D1001 were cultured in the presence or absence of HMSCs for 24 hrs and then treated with DMSO, 100nM ABT-737, 50μM MNNG, 200nM ABT-888 or combinations, as shown. The cells were stained with FITC-conjugated Annexin V, propidium iodide and pacific blue-conjugated CD19 antibodies. CD19-positive cells were gated and cells negative for both PI and Annexin V were considered alive. D. Percent of cells alive under various treatment conditions for all 16 CLL samples. Bars indicate mean and 95% confidence interval. * indicates p<0.05.

Figure 7. Model for PARP1 involvement in DNA repair and cell death

A. MDA-MB-436 breast cancer cells were transduced with BCL2 or control virus and selected in neomycin in the presence or absence of increasing concentrations of ABT-737. The experiment was performed in triplicate. B. Sensitivity of lung cancer cell lines to the PARP1 inhibitor AG014699. A replicate experiment is shown in Figure S3. C. Lung cancer cells were transduced and selected, as in (A). Representative data of three independent experiments is shown. D. Basal DNA damage from endogenous processes like replication requires PARP1 for repair. In the absence of functional PARP1, single-strand lesions can become double-strand breaks that require homology-directed repair (HDR) for resolution. In cells with impaired HDR, the accumulation of DNA double-strand breaks induces apoptosis. Similarly, extensive DNA damage from irradiation or treatment with alkylating agents can induce apoptosis, which is suppressed by BCL2. The same DNA damage can initiate repair and non-apoptotic cell death dependent on PARP1. The latter pathway is also suppressed by BCL2.