The AML cell lines MV4-11, MOLM14 and OCI-AML3 were obtained from Dr. Michael Andreeff at the University of Texas MD Anderson Cancer Center (UTMDACC). The busulfan-resistant KBM3/Bu250⁶ (KBu) cell line was established in our laboratory by serial exposure of KBM3 cells (33) to increasing concentrations of Bu. All cells were cultured in RPMI 1640 (Mediatech, Inc., Manassas, VA) with 10% heat-inactivated fetal bovine serum (FBS: Gemini Bio Products, West Sacramento, CA), 100 U/ml penicillin and 100 μg/ml streptomycin (Mediatech) at 37°C in a fully humidified atmosphere of 5% CO₂ in air.

Leukemia cell samples were isolated from patients’ peripheral blood using lymphocyte separation medium (Mediatech) and incubated in suspension in the RPMI 1640 medium described above. Patient 1 had T-cell prolymphocytic leukemia (T-PLL), patient 2 had mixed phenotype acute leukemia and patient 3 had AML, as shown in Figure 1 . Patient samples were collected after obtaining written informed consent, and all studies using these samples were performed under a protocol approved by the Institutional Review Board of the UTMDACC, in accordance with the Declaration of Helsinki.

Fludarabine (Flu), clofarabine (Clo), SAHA/vorinostat, olaparib (Ola), NU7741, LTURM34 and AZD7648 were purchased from Selleck Chemicals LLC (Houston, TX), and busulfan (Bu) was obtained from Sigma-Aldrich Chemical Sciences Corporation (St. Louis, MO). Z-VAD-FMK was purchased from Cayman Chemical Co. (Ann Arbor, MI). Drug stock solutions were dissolved in dimethyl sulfoxide (DMSO) and the final concentration of DMSO was <0.08% by volume, a level that does not induce differentiation of these cell lines.

Cell proliferation assays were done in triplicate using 96-well plates. Cell aliquots (100 µl of 0.5 X 10⁶ cells/ml) were analyzed for proliferation using the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assay as previously described (34). The inhibition of cell proliferation after a 48-h drug exposure was determined relative to the control cells exposed to solvent alone. Graphical analyses including calculations of IC₁₀–IC₂₀ values (the concentration of drug required for 10–20% growth inhibition) were done using Prism 5 (GraphPad Software, San Diego, CA). Programmed cell death was determined by flow cytometric measurements of phosphatidylserine externalization with Annexin-V-FLUOS (Roche Diagnostics, Indianapolis, IN) and 7-aminoactinomycin D (BD Biosciences, San Jose, CA) using a Muse Cell Analyzer (MilliporeSigma, St. Louis, MO).

Drug combination effects were estimated based on the combination index (CI) values calculated using the CompuSyn software (Combo Syn, Inc., Paramus, NJ). This program was developed based on the median-effect method: CI < 1 indicates synergy, CI ≈ 1 is additive, and CI > 1 suggests antagonism.

Cells were exposed continuously to drug(s) for 48 h, harvested and washed with cold phosphate-buffered saline (PBS). Cells were lysed with lysis buffer (Cell Signaling Technology, Danvers, MA). Total protein concentrations in the cell lysates were determined using a BCA Protein Assay kit (Thermo Fisher Scientific, Rockford, IL). Western blot analysis was done by separating protein extracts on polyacrylamide-SDS gels and blotting onto nitrocellulose membranes (Bio-Rad, Hercules, CA). Immunoblot analyses were done using the Immobilon Western Chemiluminescent HRP Substrate (MilliporeSigma). The sources of the antibodies and their optimum dilutions are available upon request. The β-actin protein was used as an internal control.

The levels of total PARylated proteins were determined by Western blot analysis (as described above) and enzyme-linked immunosorbent assay (ELISA) using the poly(ADP-ribose) ELISA kit from Cell Biolabs, Inc. (San Diego, CA). The monoclonal anti-PAR antibody used for Western blotting was obtained from R&D Systems, Inc. (Minneapolis, MN). The antibody is specific for PAR polymers 2 to 50 units long, but does not recognize structurally related RNA, DNA, ADP-ribose monomers, NAD, or other nucleic acid monomers.

Cells were analyzed for production of ROS using CM-H2DCFDA (5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate, acetyl ester), an ROS indicator that diffuses into cells where it is oxidized to a fluorescent product (Thermo Fisher Scientific). Briefly, cells were aliquoted (0.4 ml) into tubes and CM-H2DCFDA (3 μl of 0.12 mM solution in DMSO) was added. Cells were incubated at 37°C for 1 h and immediately analyzed with a Gallios Flow Cytometer (Beckman Coulter, Inc., Indianapolis, IN) using excitation/emission wavelengths of 492/520 nm. Arithmetic means of the fluorescence intensities were compared and the relative fold increase in ROS production was calculated.

An MMP kit (Cayman Chemical Co.) was used to determine changes in the MMP using the JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide) reagent. Cells to be analyzed were aliquoted (0.4 ml) into flow cytometry tubes. Diluted (1:10 with cell growth medium, 3 μl) MMP-sensitive fluorescent dye JC-1 was added to each tube, incubated at 37°C for 20 min, and analyzed by flow cytometry (λex = 488 nm) using the 530 nm (FL-1 channel, green) and 585 nm (FL-2 channel, red) band-pass filters. Healthy cells with functional mitochondria and high MMP exhibit red fluorescence (aggregated JC-1), whereas cells with disrupted mitochondria and low MMP show green fluorescence (monomeric JC-1).

Results are presented as the mean ± standard deviation of at least three independent experiments and statistical analysis was performed using Student’s paired t-test with a two-tailed distribution (Microsoft Excel, Redmond, WA, USA). A P value of <0.05 was considered statistically significant.

Cells were exposed to individual drugs to calculate their IC_10-20 values, which were used in the drug combination experiments to determine their effects on cell proliferation and apoptosis. Exposure of KBu, MV4-11, MOLM14, and OCI-AML3 cells to Flu, Clo, Bu, SAHA or Ola resulted in 9-28% growth inhibition. [Flu+Clo+Bu] inhibited growth by 42%, 45%, 52%, and 56% in KBu, MV4-11, MOLM14, and OCI-AML3 cells, respectively ( Figure 2A ). Addition of SAHA and Ola to the three-drug combinations inhibited cell proliferation by 66%-71% versus control ( Figure 2A ). The [Flu+Clo+Bu+Ola] combination decreased cell proliferation more than [Flu+Clo+Bu+SAHA] in AML cells (data not shown). All of these results correlate with the corresponding Annexin V data. The [Flu+Clo+Bu] combination caused ~28%-56% cell death as indicated by Annexin V positivity. Exposure to the [Flu+Clo+Bu+SAHA+Ola] combination increased Annexin V positivity to 38%-72% of control ( Figure 2A ). To test for a synergistic interaction, cells were exposed to different concentrations of the individual drugs or to the five-drug combination at a constant concentration ratio, and the MTT assay was performed. The calculated CI at increasing drug effects was graphically analyzed. Figure 2B shows combination index (CI) values <1 at fraction affected (Fa) > 0.5 in all cell lines tested, suggesting synergism for drug concentrations where potentially clinically relevant cytotoxicity was apparent. The graphs also suggest that drug antagonism (CI > 1) may exist at lower drug concentrations where lower Fa values were observed, and which resulted in relatively little cytotoxicity ( Figure 2B ).

SAHA is known to inhibit histone deacetylases, and Ola is a potent inhibitor of PARP. We, therefore, sought to determine their effects on the acetylation of histone 3 and α-tubulin as well as on the PARylation of proteins. Exposure of AML cells to low levels of SAHA (0.4 µM – 0.8 µM) did not cause any significant increase in the level of AcH3K9 but resulted in a modest increase in the level of Ac α-tubulin K40 ( Figure 2C ). The [Flu+Clo+Bu] combination had minimal effect on the acetylation of histone 3 and α-tubulin; addition of [SAHA+Ola] to this combination increased the acetylation of both proteins ( Figure 2C ). Olaparib alone at the concentrations used here (4.5 µM – 9 µM) did not decrease the PARylation levels except in MV4-11 cells; its addition to [Flu+Clo+Bu+SAHA] combination, however, markedly decreased the levels of PAR in all four AML cell lines tested ( Figure 2C ). These observations are consistent with the previously reported synergistic interactions between HDAC and PARP inhibitors (27, 30, 31); addition of [Flu+Clo+Bu] apparently enhances the inhibition of PARylation.

Whether the observed increases in Annexin V positivity ( Figure 2A ) might be associated with apoptotic cell death was assessed by analyzing the cleavage of Caspase 3 and PARP1, which are indicators of apoptosis. While [Flu+Clo+Bu] caused cleavage of Caspase 3 and PARP1, addition of SAHA and Ola to this triple-drug combination markedly enhanced their cleavage ( Figure 3A ). Phosphorylation of histone 2AX also increased in cells exposed to [Flu+Clo+Bu] and even more so after [Flu+Clo+Bu+SAHA+Ola], suggesting DSB formation and/or activation of the DNA-damage response, consistent with the observed cleavage of genomic DNA as shown by agarose gel analysis ( Figure 3B ), which was probably due to caspase-mediated activation of nuclear DNases (35).

To further identify possible mechanisms of apoptosis activation, we examined the production of ROS in AML cells exposed to [Flu+Clo+Bu+SAHA+Ola]. Flow cytometric analysis showed an ~2-fold increase in ROS levels in KBu and MV4-11 cells exposed to [Flu+Clo+Bu] versus control cells, which further increased to ~3.5- to 4-fold versus control cells after exposure to the [Flu+Clo+Bu+SAHA+Ola] 5-drug combination (P < 0.05 in both cell lines); in contrast, no significant change (P > 0.05 in both cell lines) was observed when SAHA and Ola were added to the [Flu+Clo+Bu] combination in MOLM14 and OCI-AML3 cells ( Figure 3C ). Mitochondrial membrane potential (MMP) decreased in all four cell lines after exposure to [Flu+Clo+Bu+SAHA+Ola] whereas more modest effects were seen with the [Flu+Clo+Bu] combination, as suggested by the relative increases in monomer JC-1/aggregate JC-1 ratio ( Figure 3D ); the increase in the JC-1 ratio between [Flu+Clo+Bu] and [Flu+Clo+Bu+SAHA+Ola] treatments was statistically significant (P < 0.05) in KBu and MV4-11 cell lines. The decrease in MMP may cause pro-apoptotic factors to leak from the mitochondria and activate caspases. Indeed, the level of active Caspase 3 (based on its cleavage) also increased more in cells exposed to [Flu+Clo+Bu+SAHA+Ola] than to [Flu+Clo+Bu] ( Figure 3A ). This Caspase 3 activation and decreased MMP closely paralleled the triggering of apoptosis as shown by the Annexin V assay ( Figure 2A ). Overall, cell exposure to [Flu+Clo+Bu+SAHA+Ola] induced ROS production and decreased MMP, potentially contributing to the leakage of pro-apoptotic factors and activation of caspases, and committing cells to apoptosis as shown by the cleavage of Caspase 3 and PARP1, as well as by genomic DNA fragmentation ( Figures 3A, B ). The conclusion that ROS production and mitochondrial dysfunction are in part responsible for apoptosis induction was supported by the observation that treating KBu or MV4-11 cells with 5 mM N-acetylcysteine, an ROS scavenger, for 1 h prior to the 5-drug combination treatment did not completely inhibit apoptosis as shown by partial decrease in the levels of cleaved caspase 3 and γ-H2AX ( Figure 3E ).

Since Caspase 3 was activated by the [Flu+Clo+Bu+SAHA+Ola] combination, we wished to determine if the observed inhibition of PARylation was caspase-dependent. To this effect, cells were exposed to an irreversible pan-caspase inhibitor, Z-VAD-FMK, 30 min prior to addition of the five-drug combination and analyzed by Western blotting and ELISA. Figure 4A shows PARylation of multiple proteins in both the control and [Flu+Clo+Bu] treatment groups. The [SAHA+Ola] and [Flu+Clo+Bu+SAHA+Ola] combinations significantly inhibited PARylation; however, addition of Z-VAD-FMK did not affect this drug-mediated inhibition ( Figure 4A ). Again, significant cleavage of Caspase 3 and PARP1 was observed in cells exposed to [Flu+Clo+Bu+SAHA+Ola] and addition of Z-VAD-FMK inhibited Caspase 3 cleavage in MV4-11 and MOLM14 cells but had little effect on PARP1 cleavage ( Figure 4A ). Interestingly, the phosphorylation of histone 2AX seen after [Flu+Clo+Bu+SAHA+Ola] treatment was markedly inhibited by Z-VAD-FMK in MV4-11 cells but not in MOLM14 cells ( Figure 4A ). Quantitative determination of the levels of PARylation by ELISA provided similar results ( Figure 4B ), suggesting that the observed inhibition of PARylation is caspase independent.

To assess the potential clinical extension of our results, cells from patients with leukemia were exposed to [Flu+Clo+Bu], [SAHA+Ola], or [Flu+Clo+Bu+SAHA+Ola] and analyzed by Western blotting. Increased cleavage of Caspase 3 was apparent in patient samples 1 and 3 exposed to the five-drug combination (patient 2 had protein loading problems as indicated by unequal levels of β-ACTIN), suggesting robust activation of the apoptosis pathway; a modest cleavage of PARP1 was observed in cells exposed to the 5-drug combination ( Figure 1 ). Phosphorylation of histone 2AX was also observed in cells exposed to the 5-drug combination, suggesting DSB formation and/or activation of the DNA-damage response. These findings suggest a synergism of [Flu+Clo+Bu+SAHA+Ola] in cells derived from patients with leukemia involving mechanisms similar to those seen in the cultured cell lines. Additional patient samples may need to be analyzed to confirm the clinical relevance of these results.

Acetylation and PARylation are known to occur in some proteins involved in DNA repair (27, 36). These post-translational modifications affect the stability of the proteins, as previously shown for UHFR1 and BRCA1 (37, 38). We, therefore, examined the effects of Flu, Clo, Bu, SAHA and Ola, individually or in combinations, on DNA damage response protein levels (total and phosphorylated) in AML cell lines. In general, the levels of phosphorylated and pan ATM (which functions in DNA DSB repair and cell cycle checkpoint activation) decreased in cells exposed to the five-drug combination; the level of BRCA1 (which functions in homologous recombination (HR) repair) greatly decreased ( Figure 5 ). ATRX is a chromatin remodeling protein involved in HR (39) while DAXX has multiple functions in human cells, including regulation of DNA repair (40). The levels of both proteins decreased in AML cells exposed to the five-drug combination ( Figure 5 ). While the level of the non-homologous end joining (NHEJ) repair protein DNAPKcs decreased in cells exposed to [Flu+Clo+Bu+SAHA+Ola], its phosphorylation at serine 2056 dramatically increased; the level of another NHEJ protein, Artemis, also decreased ( Figure 5 ).

The NuRD complex is involved in chromatin remodeling and deacetylation processes (41) and plays a key role in the cellular DNA damage response by regulating DNA damage signaling and repair events (42). The levels of the CHD3, CHD4, RBAP46, and MTA1 subunits of NuRD decreased in all cell lines exposed to [Flu+Clo+Bu+SAHA+Ola]; the HDAC1 subunit also modestly decreased, but HDAC2 was only minimally affected ( Figure 5 ).

Of all the DNA damage response/repair protein events analyzed, the increased phosphorylation of DNAPKcs at serine 2056 was most intriguing. We, therefore, analyzed the kinetics of its phosphorylation. Figure 6A shows a marked increase in the level of P-DNAPKcs (S2056) within 6 h of exposure to [Flu+Clo+Bu+SAHA+Ola], similar to the kinetics of P53 phosphorylation at serine 15 and phosphorylation of histone 2AX in the MV4-11 and MOLM14 cell lines. These results are consistent with the reported function of DNAPKcs, which is to control the repair of the broken DNA ends and transmit the damage signal through P53 protein to induce cell cycle arrest and apoptosis (43). Whereas the level of P-DNAPKcs increased with drug treatment, the level of unphosphorylated DNAPKcs decreased within 24 h, which coincides with the cleavage of Caspase 3. DNAPKcs autophosphorylation at Ser2056 results in inactivation of its kinase activity and DNA repair ability (44–46). These results suggest significant effects of [Flu+Clo+Bu+SAHA+Ola] on the complex interactions among proteins involved in the DNA damage response, DNA repair, cell cycle, and programmed cell death.

To further analyze the possible interaction between DNAPK and P53, MV4-11 cells were exposed to DNAPK inhibitors prior to addition of the chemotherapy drugs. The inhibitors Nu7741, LTURM34 and AZD7648 increased the level of P-DNAPKcs (S2056) but decreased the level of pan P53 relative to the control (i.e., no inhibitor) after the 24 h five-drug exposure ( Figure 6B ), suggesting that the stability of P53 was compromised in the presence of DNAPK inhibitors. These findings are consistent with DNA-PK phosphorylating and stabilizing P53 in this setting as previously reported (47, 48).