53BP1 depletion causes PARP inhibitor resistance in ATM-deficient breast cancer cells

Breast cancer cell lines CAL-51 and ZR-75-1 were cultured in RPMI 1640 (Gibco) with 10 % FBS. Cell lines MCF-7, T47D, MDA-MB-231, MDA-MB-468 and SK-BR-3 were cultured in DMEM (Gibco) containing 10 % FBS. All cultures were maintained at 37 °C in a 5 % CO₂ atmosphere.

Lentivirus encoding the short hairpin RNAs shRNA1 and shRNA2 targeting 53BP1 and the non-specific target shRNA were designed and prepared by Genechem (Shanghai, China). CAL-51 and MCF-7 cells were infected with a pool of lentiviruses carrying shRNA1 or shRNA2 and selected with 1 μg/mL puromycin for 10 days.

The PARP-1 inhibitor Olaparib (AZD2281) and ATM inhibitor KU55933 were purchased from Selleck Chemicals. Compounds were serially diluted in DMSO and finally diluted 10³-fold in culture medium immediately before use.

Cells were seeded in 96-well plates at the appropriate density for each cell line in 100 μl of medium and left at room temperature for 30 min, after which they were incubated overnight at 37 °C to allow attachment. Then cells were incubated for 48 h at 37 °C with Olaparib at concentrations of 0, 1, 2.5, 5, or 10 μM in the presence or absence of 10 μM KU55933. During incubation with inhibitors, the medium was not changed, nor were inhibitors added again.

Then 10 μl of MTS reagent (Promega, USA) was added directly to the wells, and plates were incubated at 37 °C for at least 1 h. Absorbance was measured at 490 nm on a microplate reader (Bio-Rad 680, USA). Background absorbance was first subtracted using a set of wells containing medium only, then normalized to and expressed as a relative percentage of the plate-averaged DMSO control.

Cells (5 x 10³) were seeded into 6-well culture plates and incubated for 10 days at 37 °C in the presence of Olaparib at 0, 2.5, 5, or 10 μM in the presence or absence of 10 μM KU55933. Cells were washed with pre-chilled phosphate-buffered saline, fixed for 20 min with pre-chilled methanol and stained for 15 min with crystal violet. Colonies were examined and automatically counted using G:box (SYNGENE). All experiments were performed in triplicate.

Human breast cancer cell lines CAL-51 and MCF-7 were treated for 48 h with DMSO vehicle, 10 μM KU55933, 10 μM Olaparib, or the combination of 10 μM Olaparib and 10 μM KU55933. Cells were harvested by trypsinization, resuspended to a density of 1 × 10⁶ cells/mL in 1 × binding buffer, and stained with FITC-Annexin V and propidium iodide (PI) using the FITC Annexin V Apoptosis Detection Kit I (BD Biosciences). Stained cells were analyzed using an LSRII flow cytometer (Falcon BD, San Jose, CA, United States) according to the manufacturer’s instructions. All experiments were performed in triplicate.

Cells were harvested and lysed on ice for 40 min in RIPA buffer (10 mM Tris pH 7.4, 150 mM NaCl, 1 % Triton X-100, 0.1 % Na-deoxycholate, 0.1 % SDS and 5 mM EDTA) containing Complete Protease Inhibitor Cocktail (Roche Applied Science). Total protein concentration was determined using a colorimetric assay. Equal amounts of total protein (60 μg) were separated by 6 %, 8 %, or 10 % SDS-PAGE and transferred to PVDF membranes. Membranes were blocked with 2 % bovine serum albumin (BSA), then incubated with primary antibodies overnight at 4 °C. Primary antibodies from Cell Signaling (Boston, MA, USA) were against caspase 3, cleaved caspase 3, caspase 8, cleaved caspase 8, caspase 9, cleaved caspase 9, PARP-1, cleaved PARP-1, Chk2, phospho-Chk2 (Thr68), phospho-ATM (Ser1981) and γ-H2AX. Primary antibodies from Abcam were against ATM and Rad51. Primary antibody against β-actin (Sigma) provided a loading control. The membrane was then incubated for 1 h with horseradish peroxidase-conjugated goat anti-mouse IgG (1:2000) or goat anti-rabbit IgG (1:3000) secondary antibody. The membrane was rinsed in 1 × PBS containing 0.1 % Tween, then incubated with chemiluminescence substrate. Blots were photographed using an Image Reader LAS-4000 (Fujifilm) and analyzed using Multi Gauge 3.2.

CAL-51 and MCF-7 cells were treated with DMSO, 10 μM Olaparib with or without 10 μM KU55933 respectively for 48 h, washed three times with PBS, fixed with methanol, and permeabilized with 0.1 % Triton X-100. Then cells were stained with DAPI at 37 °C for 5 min in the dark. Stained cells were washed twice with PBS before observed under fluorescence microscopy.

CAL-51 and MCF-7 cells were treated for 48 h with DMSO vehicle or with Olaparib (5 or 10 μM) in the presence or absence of 10 μM KU55933, then fixed with 4 % paraformaldehyde. Fixed cells were incubated for 1 h in 1 % BSA/10 % normal goat serum/0.3 M glycine containing 0.05 % PBS-Tween in order to permeabilize the cells and block non-specific protein–protein interactions. Cells were incubated with anti-γ-H2AX antibody (Cell Signaling), followed by Alexa Fluor® 488-conjugated secondary antibody (ZSGB-BIO, Beijing, China). Nuclei were stained using DAPI. Cells were visualized using a non-confocal Operetta automated microscope (Perkin Elmer). The number of γ-H2AX foci per cell was quantified by ImageJ software. At least 200 cells per experiment point were examined.

Formalin-fixed, paraffin-embedded (FFPE) tumor samples were obtained from breast cancer patients treated at the Cancer Institute and Hospital of the Chinese Academy of Medical Sciences and Peking Union Medical College. Clinical and follow-up data on the patients were extracted from the clinical database. The use of clinical specimens and data was approved by the local ethics committee of the Cancer Institute and Hospital of Peking Union Medical College.

FFPE samples of triple-negative breast cancer were immunohistochemically stained with a monoclonal antibody against phospho-ATM (1:100; rabbit IgG; Epitomics, clone EP1890Y, S1981). This antibody binds specifically to ATM phosphorylated Ser1981 [25]. Samples were stained for 53BP1 using a rabbit polyclonal antibody against 53BP1 (1:100; rabbit IgG; Santa Cruz Biotechnology, clone H-300, catalog no. SC-22760). Slides were deparaffinized with xylene and rehydrated in graded ethanol. Sections were submerged in EDTA antigenic retrieval buffer (pH 8.0), microwaved to retrieve antigens, treated with 3 % hydrogen peroxide in methanol to quench endogenous peroxidase activity, and finally incubated with 1 % goat serum albumin to block non-specific binding. Sections were incubated overnight at 4 °C with antibodies against phospho-ATM or 53BP1, washed, then treated for 40 min with goat anti-mouse or -rabbit IgG conjugated to horseradish peroxidase (ZSGB-BIO). The chromogen was 3,3′-diaminobenzidine.

Statistical analysis was carried out using SPSS 20 (IBM, Chicago, IL, USA) or GraphPad Prism 5 for Windows, with a significance threshold of p < 0.05. Results were expressed as mean ± SEM, and differences were assessed for significance using the two-tailed Student’s t test. The two-tailed Pearson χ² test or Fisher’s exact test was used to identify associations of phospho-ATM or 53BP1 expression with clinicopathological parameters. Survival curves were plotted using Kaplan-Meier analysis and compared using the log-rank test. Factors associated with survival were identified using multivariate Cox regression.

To evaluate whether ATM inhibition increases the cytotoxicity of PARP inhibitors, we treated both triple-negative breast cancer cell lines (CAL-51, MDA-MB-231 and MDA-MB-468) and non-triple-negative breast cancer cell lines (MCF-7, T-47D and SK-BR-3) with increasing concentrations of Olaparib (0–10 μM) alone or in combination with 10 μM ATM inhibitor KU55933. All cell lines had wild-type BRCA1 and BRCA2 genes; Additional file 1: Table S1 summarizes the ER, PR, HER2, BRCA1 and BRCA2 status of the lines. After 48-h incubation with one or both inhibitors, cell viability was assessed using the MTS assay. Olaparib induced ATM phosphorylation at Ser1981, which was substantially inhibited in the presence of KU55933 (Fig. 1a). This suggests that ATM activation participates in the cellular response to PARP inhibition. KU55933 significantly increased Olaparib toxicity in the triple-negative cell line CAL-51 and non-triple-negative cell line MCF-7 (Fig. 1b-c). Similar results were observed in the other cell lines (Additional file 1: Figure S1A). Interestingly, the three triple-negative cell lines appeared to be more sensitive to Olaparib in the presence of KU55933 than the three non-triple-negative cell lines. Olaparib inhibited colony formation in CAL-51 and MCF-7 cells in a dose-dependent manner, which KU55933 further enhanced it (Fig. 1d-e). Taken together, these data suggest that Olaparib suppresses cell proliferation and colony formation in breast cancer cell lines, and that KU55933 reinforces this effect through synthetic lethality.

An early cellular response to DSBs is rapid phosphorylation of H2AX at Ser139, forming γ-H2AX foci [26]. H2AX phosphorylation plays a key role in signalling and initiating the repair of DSBs, and the numbers of γ-H2AX foci correlate closely with the number of DSBs [27]. Treating breast cancer cell lines with both Olaparib and KU55933 led to significantly more γ-H2AX foci formation than treating them with either inhibitor alone based on fluorescence microscopy (Fig. 2a-b). These findings were further confirmed by Western Blot assay (Fig. 2c). This provides further evidence that that Olaparib and Ku-55933 synthetically induce cellular DSBs.

Since the experiments above indicated that Olaparib and KU55933 increased DSB formation, we wanted to know whether they induced changes in nuclear morphology. Such changes could reflect DSB-induced chromosome rearrangements or mutations, which can lead to apoptosis or death [23]. Treating MCF-7 and CAL-51 cells for 48 h with 10 μM Olaparib, 10 μM KU55933 or both led to nuclei with chromatin condensation and apoptotic bodies detectable by DAPI fluorescence (Fig. 3a-b), with the combination treatment causing more severe changes in nuclear morphology than either inhibitor alone. Similarly, combination treatment caused significantly greater apoptosis than Olaparib alone, based on Annexin V/PI staining followed by fluorescence-activated cell sorting (Fig. 3c-d). In addition, Western blotting experiments showed that combination treatment, as well as KU55933 treatment on its own, up-regulated expression of apoptotic caspases 3, 8, and 9, and their cleaved forms and it stimulated PARP-1 cleavage (Fig. 3e). These results suggest that Olarparib and KU55933 activate apoptosis pathways.

The 53BP1 gene was stably knocked-down by transfecting CAL-51 and MCF-7 cell lines with sh53BP1 lentivirus particles. Particles were generated using two siRNAs to ensure efficient knock-down. Stable transfectants, named MCF-7-sh53BP1 and CAL-51- sh53BP1, were selected in the presence of puromycin for 10 days and maintained as polyclonal populations. Both transfected cell lines showed much lower 53BP1 expression than the control-transfected lines (Fig. 4a). Both transfected lines were also significantly less sensitive than control-transfected lines to 10 μM Olaparib, 10 μM KU55933 or both (Fig. 4b-c). Specifically, following combination treatment, proliferation inhibition rates were 20.3 % for MCF-7-sh53BP1 cells, 29.8 % for MCF-7-ctrl cells, 46.5 % for CAL-51-sh53BP1 and 57.4 % for CAL-51-ctrl cells (Additional file 1: Figure S2). Similar results were obtained in the colony formation assay (Fig. 4d-e). Taken together, our results suggest that 53BP1 deficiency renders ATM-inhibited cells less sensitive to Olaparib.

It is well accepted that ATM, Chk2 and Rad51 are critical regulators of HR DNA repair pathway. We sought to investigate the protein levels of these factors under the treatment with Olaparib, KU55933 or both. As is shown in Fig. 5a, Olaparib induced ATM phosphorylation and activated its downstream effector Chk2, and subsequently up-regulated Rad51. Adding KU55933 inhibited the activation of ATM and Chk2 and down-regulated Rad51. Taken together, these observations suggest that Olaparib activated the HR pathway and adding KU55933 inhibited it.

Previous studies have reported that knockdown of 53BP1 in BRCA1 deficient cells partially restored HR activity [22, 23]. Interestingly, we found that knock-down of 53BP1 also restored HR activity in ATM deficient cells. Higher levels of phospho-ATM and Chk2 and up-regulated RAD51 were observed in 53BP1 knock-down cells than their control counterparts when treated with Olaparib and KU55933 combined (Fig. 5b).

Our experiments link levels of phospho-ATM and 53BP1 to Olaparib sensitivity of breast cancer cell lines. Further, we examined whether these levels might correlate with survival of triple-negative breast cancer patients. Immunohistochemistry of FFPE tumor samples from 73 patients with early triple-negative breast cancer identified low expression levels of phospho-ATM in 44 (60.3 %) and high levels in 29 (39.7 %). Overall survival was significantly better in patients with low levels of phospho-ATM (Fig. 6a). After taking into account various clinical characteristics of the patients (Additional file 1: Table S2), and adjusting for age, tumor size, grade, number of metastatic lymph nodes and lymphovascular invasion, multivariate analysis identified phospho-ATM level as an independent predictor of overall survival (HR 1.83, 95%CI 1.063–3.164, P = 0.029; Fig. 6b).

Immunohistochemistry of FFPE tumor samples from another cohort of 92 patients with triple-negative breast cancer identified 67 (72.8 %) as negative for 53BP1. Overall survival tended to be better among these patients than among patients positive for 53BP1, but the difference did not achieve significance (Fig. 6c). The clinicopathological characteristics of this cohort are shown in Additional file 1: Table S3.)