An Optically Pure Apogossypolone Derivative as Potent Pan-Active Inhibitor of Anti-Apoptotic Bcl-2 Family Proteins

Our focus in the past several years has been on the identification of novel and effective pan-Bcl-2 antagonists. We have recently reported a series of Apogossypolone (ApoG2) derivatives, resulting in the chiral compound (±) BI97D6. We report here the synthesis and evaluation on its optically pure (−) and (+) atropisomers. Compound (−) BI97D6 potently inhibits the binding of BH3 peptides to Bcl-XL, Bcl-2, Mcl-1, and Bfl-1 with IC50 values of 76 ± 5, 31 ± 2, 25 ± 8, and 122 ± 28 nM, respectively. In a cellular assay, compound (−) BI97D6 effectively inhibits cell growth in the PC-3 human prostate cancer and H23 human lung cancer cell lines with EC50 values of 0.22 ± 0.08 and 0.14 ± 0.02 μM, respectively. Similarly, compound (−) BI97D6 effectively induces apoptosis in the BP3 human lymphoma cell line in a dose-dependent manner. The compound also shows little cytotoxicity against bax−/−/bak−/− cells, suggesting that it kills cancers cells predominantly via a Bcl-2 pathway. Moreover, compound (−) BI97D6 displays in vivo efficacy in both a Bcl-2-transgenic mouse model and in a prostate cancer xenograft model in mice. Therefore, compound (−) BI97D6 represents a promising drug lead for the development of novel apoptosis-based therapies for cancer.


INTRODUCTION
Programmed cell death (apoptosis; Reed, 1999;Vaux and Korsmeyer, 1999) plays critical roles in both the onset and progress of cancer and contributes significantly to chemoresistance (Johnstone et al., 2002;Reed, 2002). B-cell lymphoma/leukemia-2 (Bcl-2) family proteins are central regulators of the apoptotic machinery (Adams and Cory, 1998;Reed, 1998;Gross et al., 1999). The Bcl-2 family is composed of anti-apoptotic and proapoptotic members. To date, six anti-apoptotic members of the Bcl-2 family have been identified and characterized, including Bcl-2, Bcl-X L , Mcl-1, Bfl-1, Bcl-W, and Bcl-B. Since the overexpression of anti-apoptotic Bcl-2 family proteins is associated with tumor progression, poor prognosis, and drug resistance, these proteins are representing attractive targets for anticancer drug (Reed, 1997;Wang et al., 2000;Degterev et al., 2001). X-ray crystallography and nuclear magnetic resonance (NMR) spectroscopy structural studies have elucidated a hydrophobic crevice on the surface of anti-apoptotic Bcl-2 family proteins that binds the BH3 dimerization domain of pro-apoptotic family members (Muchmore et al., 1996;Sattler et al., 1997;Reed, 1998). Thus, molecules that mimic the BH3 domain of pro-apoptotic proteins may be effective in either inducing apoptosis and/or in abrogating the ability of antiapoptotic Bcl-2 proteins to inhibit cancer cell death (Reed, 1997(Reed, , 1998Wang et al., 2000;Degterev et al., 2001).
Indeed, the (−) Gossypol displayed a marked differential activity compared to its natural racemic mixture . Therefore, in this current work, we focus our attention on preparing pure atropisomers of (±) ApoG2 and its most potent derivative (BI97D6) followed by further investigation of their in vitro and in vivo activities ( Figure 1C).

MOLECULAR MODELING
Molecular modeling studies were conducted on a Linux workstation and a 64 3.2-GHz CPUs Linux cluster. Docking studies were performed using the crystal structure of Bcl-X L in complex with a BH3 mimetic ligand (Protein Data Bank code 2YXJ; Oltersdorf et al., 2005;Bruncko et al., 2007;Lee et al., 2007). The ligand was extracted from the protein structure and was used to define the binding site for small molecules. Compounds (−) and (+) BI97D6 were docked into the Bcl-X L protein by the GOLD (Jones et al., 1997) docking program using ChemScore (Eldridge et al., 1997) as the scoring function. The active site radius was set at 10 Å and 10 GA solutions were generated for each molecule. The GA docking procedure in GOLD (Jones et al., 1997) allowed the small molecules to flexibly explore the best binding conformations whereas the protein structure was static. The protein surface was prepared with the program MOLCAD (Teschner et al., 1994) as implemented in Sybyl (Tripos, St. Louis) and was used to analyze the binding poses for studied small molecules.

NMR EXPERIMENTS
Nuclear magnetic resonance-based binding assays have been conducted by acquiring one-dimensional 1 H experiments with 500 μL solution of Bcl-X L at 20 μM concentration, in absence and presence of added compounds, each at 100 μM concentration. By observing the aliphatic region of the spectra, binding could be readily detected due to chemical shift changes in active site methyl groups of Ile, Leu, Thr, Val, or Ala (region between −0.8 and 0.3 ppm). The Mcl-1 binding mode was characterized by recording [ 15 N, 1 H]-HSQC experiments with 500 μL solution of uniformly 15 N-labeled Mcl-1 (25 μM concentration) in absence and presence of added compounds, each at 25 and 125 μM concentration, respectively. Bcl-X L and Mcl-1 samples were prepared and purified as described previously (Sattler et al., 1997;Day et al., 2005). All experiments were performed with a 600-MHz spectrometer Bruker Advance 600 equipped with four rf channels and z-axis pulse-field gradients.

FLUORESCENCE POLARIZATION ASSAYS
A Bak BH3 peptide (F-BakBH3) (GQVGRQLAIIGDDINR) was labeled at the NH 2 terminus with fluorescein isothiocyanate (FITC; Molecular Probes) and purified by HPLC. For competitive binding assays, 20 nM GST-Bcl-X L ΔTM protein was preincubated with the tested compound at varying concentrations in 47.5 μL phosphate-buffered saline (PBS; pH = 7.4) in 96-well black plates at room temperature for 5 min, then 2.5 μL of 15 nM FITC-labeled Bim BH3 (FITC-Ahx-DMRPEIWIAQELRRIGDEFNAYYAR) peptide was added to produce a final volume of 50 μL. The wild-type and mutant Bim BH3 peptides were included in each assay plate as positive and negative controls, respectively. After 10 min incubation at room temperature, the polarization values in millipolarization units (Sattler et al., 1997) were measured at excitation/emission wavelengths of 480/535 nm with a multilabel plate reader (PerkinElmer). IC 50 was determined by fitting the experimental data to a sigmoidal doseresponse non-linear regression model (SigmaPlot 10.0.1; Systat Software, Inc., San Jose, CA, USA). Data reported are mean of three independent experiments ± SE. Performance of Bcl-2, Mcl-1, and Bfl-1 FPA are similar. Briefly, 20 nM of GST-Bcl-2 or -Mcl-1, or -Bfl-1 were incubated with various concentrations of compounds (−), (+), (±) ApoG2 and BI97D6 for 2 min, then 15 nM FITCconjugated-Bim BH3 peptide (Ramjaun et al., 2007) was added in PBS buffer. Fluorescence polarization (FP) was measured after 10 min.

CELL VIABILITY ASSAYS
The activity of the compounds against human cancer cell lines (PC-3, H460, and H23) were assessed by using the ATP-LITE assay (PerkinElmer). All cells were seeded in either 12F2 or RPMI1640 medium with 5 mM l-glutamine supplemented with 5% fetal bovine serum (Mediatech Inc.), penicillin, and streptomycin (Omega). For maintenance, cells were cultured in 5% FBS. Cells plated into 96-well plates at varying initial densities depending on doubling time. H460 and H23 plated at 2000 cells/well and PC-3 at 3000 cells/well. Compounds were diluted to final concentrations with 0.1% DMSO. Prior to dispensing compounds onto cells, fresh 5% media was placed into wells. Administration of compounds occurred 24 h after seeding into the fresh media. Cell viability was evaluated using ATP-LITE reagent (PerkinElmer) after 72 h of treatment. Data were normalized to the DMSO control-treated cells using Prism version 5.01 (GraphPad Software). Data were reported as mean of three independent experiments ± SE. The activity of compounds (−) and (+) BI97D6 against mouse embryonic fibroblast wild-type cells (MEF/WT) and mouse embryonic fibroblast BAX/Bak double knockout cells (DKO/MEF) was assessed by 1 day ATP-LITE assay. Wild-type MEF and DKO/MEF were seeded in 96-well plate at a seeding density of 10,000 cells per well. The next day, compounds (−) and (+) BI97D6 were added to wild-type and DKO cells. Cell viability was evaluated using ATP-LITE reagent (PerkinElmer) after 24 h of treatment. Data were normalized to the DMSO control-treated cells using Prism version 5.01 (GraphPad Software). Data were reported as mean of three independent experiments ± SE.

Liver microsomal stability
Pooled rat liver microsomes (BD Biosciences, # 452701) were preincubated with test compounds at 37.5˚C for 5 min in the absence of NADPH. The reaction was initiated by addition of NADPH and then incubated under the same conditions. The final incubation concentrations were 4 μM test compound, 2 mM NADPH, and 1 mg/mL (total protein) liver microsomes in PBS at pH 7.4. One aliquot (100 μL) of the incubation mixture was withdrawn at 0, 15, 30, and 60 min and combined immediately with 200 μL of ACN/MeOH containing an internal standard. After mixing, the sample was centrifuged at approximately 13,000 rpm for 12 min. The supernatant was transferred into an auto sampler vial and the amount of test compound was quantified using the Shimadzu LCMS-2010EV mass spectrometer. The change of the AUC (area under the curve) of the parent compound as function of time was used as a measure of microsomal stability. Data reported are mean of two independent experiments ± SE.

Plasma stability
A 20-μL aliquot of a 10-mM solution in DMSO of the test compound was added to 2.0 mL of heparinized rat plasma (Lampire, P1-150N) to obtain a 100-μM final solution. The mixture was incubated for 1 h at 37.5˚C. Aliquots of 100 μL were taken (0, 30 min, 1 h) and diluted with 200 μL of MeOH containing internal standard. After mixing, the sample was centrifuged at approximately 13,000 rpm for 12 min. The supernatant was transferred into an autosampler vial and the amount of test compound was quantified using the Shimadzu LCMS-2010EV system. The change of the AUC (area under the curve) of the parent compound as function of time was used as a measure of microsomal stability. Data reported are mean of two independent experiments ± SE.

Parallel artificial membrane permeation assay
A 96-well microtiter plate (Millipore, # MSSACCEPTOR) was completely filled with aqueous buffer solution (pH 7.2) and covered with a microtiter filterplate (Millipore, # MAPBMN310). The hydrophobic filter material was impregnated with a 10% solution of hexadecane in hexane and the organic solvent was allowed to completely evaporate. Permeation studies were started by the transfer of 200 μL of a 100-μM test compound solution on top of the filterplate. In general phosphate buffer at pH 7.2 buffer was used. The maximum DMSO content of the stock solutions was <5%. In parallel, an equilibrium solution lacking a membrane was prepared using the exact concentrations and specifications but lacking the membrane. The concentrations of the acceptor and equilibrium solutions were determined using the Shimadzu LCMS-2010EV and AUC methods. The permeation of a compound through the membrane layer is described by the percentage permeation (% flux). The flux values were calculated considering the concentration of the acceptor compartment after 8 h and that of a reference well with the same concentration containing no membrane barrier.

Bcl-2-TRANSGENIC MICE STUDIES
Transgenic mice expressing Bcl-2 have been described as the B6 line (Katsumata et al., 1992). The BCL-2 transgene represents a minigene version of a t(14;18) translocation in which the human BCL-2 gene is fused with the immunoglobulin heavy-chain (IgH) locus and associated IgH enhancer. The transgene was propagated on the Balb/c background. These mice develop polyclonal B-cell hyperplasia with asynchronous transformation to monoclonal aggressive lymphomas beginning at approximately 6 months of age, with approximately 90% of mice undergoing transformation by the age of 12-24 months. All animals used here had not yet developed aggressive lymphoma. Compounds dissolved in 500 μL of solution (Ethanol:Cremophor EL:Saline = 10:10:80) were injected intraperitoneally to age-and sex-matched B6Bcl-2 mouse, while control-mice were injected intraperitoneally with 500 μL of the same formulation without compound. After 24 h, B6Bcl-2 mice were sacrificed by intraperitoneal injection of lethal dose of Avertin. Spleen was removed and weighed. The spleen weight of mice is used as an end-point for assessing activity as we determined that spleen weight is highly consistent in ageand sex-matched Bcl-2-transgenic mice in preliminary studies . Variability of spleen weight was within ±2% among control-treated age-matched, sex-matched B6Bcl-2 mice. Data reported in Figure 3C are mean of five independent experiments ± SE.

HUMAN PROSTATE CANCER XENOGRAFTS IN ATHYMIC NUDE MICE
PC-3-Luc cells (1 × 10 6 ) were injected s.c. in 100 μL of PBS in the left flank of male athymic nude mice (NCR nu/nu , 4 weeks old, 20 g body weight). After establishing visible tumors of ∼75mm 3 , requiring ∼5-6 days, compound dissolved in 500 μL of solvent (ethanol/Cremophor EL/saline = 10:10:80) were injected intraperitoneally (i.p.). The injections were given every 2 days for a total of nine injections. Seven treatment groups were established for the experiment, i.e., DMSO only, 3 mg/kg of (±) BI97D6, (−) BI97D6, (+) BI97D6, 5 mg/kg of (± ) BI97D6, (−) BI97D6, (+) BI97D6, respectively. A minimum of five animals was used per experimental condition. Data reported are mean of five independent experiments ± SE. For in vivo imaging of tumors, the mice were anesthetized and injected i.p. with 150 mg/kg luciferin and light emitted from each tumor was determined using a Xenogen system with CCD camera with an integration time of 1 min. Luminescence measurements were made using Living Image software (version 2.50.1; Xenogen).

RESULTS AND DISCUSSION
We had recently reported that (±) BI97D6 ( Figure 1B) was a promising inhibitor of Bcl-X L , Bcl-2, and Mcl-1 with improved in vitro and in vivo efficacy compared to (±) ApoG2 (Wei et al., 2010a). However, by using chiral chromatography we found that, similar to what was known about (±) Gossypol and Apogossypol Wei et al., 2009c), compound ApoG2 Frontiers in Oncology | Cancer Molecular Targets and Therapeutics and its derivative (BI97D6) also displayed axial chirality due to restricted rotation around the binaphtyl bond ( Figure 1B). Therefore, it was attractive to explore whether optically pure (−) and (+) atropisomers of ApoG2 and BI97D6 presented different in vitro and in vivo activities. In order to predict binding poses of (−) and (+) BI97D6 into the BH3 binding groove of Bcl-X L , molecular docking studies were performed (Figures 2A,B). Analysis of the predicted binding models indicated that both atropisomers (−) BI97D6 and (+) BI97D6 could fit well into the BH3 binding groove of Bcl-X L (Figures 2A,B), with the left half components of (−) and (+) atropisomers bound to Bcl-X L in a similar orientation. However, their right substituted naphthalene rings were predicted to present rather different binding modes (Figures 2A,B). The (−) BI97D6 was predicted to form hydrogen bonding with residue Asn 136 in Bcl-X L through its 1 oxygen on the right naphthalene ring (Figure 2A), whereas the (+) BI97D6 could not form the hydrogen bonding ( Figure 2B). The chemscore of (−) BI97D6 was 28.54, which was greater than 20.25 for (+) BI97D6. Therefore, we anticipated that the atropisomer (−) BI97D6 might have higher binding affinity for Bcl-X L compared to its (+) atropisomer.
Taken together, these observations suggested that unlike currently available antagonists (van de Donk et al., 2003;Oltersdorf et al., 2005), our compounds (−) BI97D6, (+) BI97D6, and (±) BI97D6 were effective in inhibiting several anti-apoptotic Bcl-2 proteins, and therefore were expected to display in vivo efficacy against a variety of in vivo models that relied on different Bcl-2 proteins for growth and progression Dash et al., 2010;Placzek et al., 2010). To test this hypothesis, we selected two different in vivo models: a Bcl-2-transgenic mouse model and a prostate cancer xenograft model Dash et al., 2010). B-cells of B6 transgenic mice overexpressed human Bcl-2 proteins and accumulated in the spleen resulting of a measurable weight enlargement. Because we had determined that the spleen weight was highly consistent in age-and sex-matched Bcl-2-transgenic mice, varying by only ±2% among control Bcl-2 mice , the spleen weight was used as an end-point for assessing in vivo activity of anti-Bcl-2 compounds . B6Bcl-2-transgenic mice were treated with compounds (±) BI97D6, (−) BI97D6, and (+) BI97D6 in doses of 5, 10, and 20 μmol/kg (i.p.), respectively ( Figure 3B). In agreement with our in vitro and cell data, compounds (±) BI97D6, (−) BI97D6, and (+) BI97D6 induced significant spleen weight reduction in mice in a dose-dependent manner. In particular, compound (−) BI97D6 induced 27, 38, and 41% spleen weight reduction in a single-dose of 5, 10, and 20 μmol/kg, respectively ( Figure 3B). Since the maximum spleen shrinkage would be no more than 50% in this experimental model , compound (−) BI97D6 induced near 54, 76, and 82% maximal biological activity. Given that compounds (−), (+), and (±) BI97D6 has comparable binding affinity against Bcl-2 in FP assays (Table 1), compounds (+) and (±) BI97D6 has similar activities as (−) BI97D6 and induced 81, 77% maximal biological activity in a single-dose of 10 μmol/kg, respectively ( Figure 3B). However, compounds (±) BI97D6, (−) BI97D6, and (+) BI97D6 also exhibited toxicity profiles that were more pronounced at higher doses. For instance, mice treated with (−) BI97D6 in a dose of 10 and 20 μmol/kg displayed mild to moderate hepato-toxicity, respectively, whereas mice treated with (−) BI97D6 in a dose of 5 μmol/kg displayed no evident sign of toxicity. To confirm the results of the single transgenic mouse experiment at a dose of 5 μmol/kg, we next evaluated the in vivo activity of compounds (±) BI97D6, (−) BI97D6, and (+) BI97D6 in groups of six B6Bcl-2-transgenic mice each at a dose of 5 μmol/kg. Consistent with the single mouse experiment, compounds (−), (±), and (+) BI97D6 treatment resulted in 47 ± 2, 45 ± 7, and 52 ± 2%, respectively, maximal reduction of spleen weight compared to the control group of six mice at a dose of 5 μmol/kg ( Figure 3C). All mice tolerated the treatment well, with no evident signs of toxicity.
As anticipated earlier, we and others had recently realized the importance of Mcl-1 inhibition in designing Bcl-2 antagonist (van de Donk et al., 2003;Oltersdorf et al., 2005;Placzek et al., 2010;Wei et al., 2010a,b;Dash et al., 2011). For example, the potent Bcl-X L /Bcl-2 antagonist ABT-737 and the Bcl-2 antisense Genasense (Genta) were not effective against cancer cells that overexpressed Mcl-1 (van de Donk et al., 2003;Oltersdorf et al., 2005;Placzek et al., 2010). Therefore, to further examine the therapeutic potential of our molecules as single agents against tumors, compounds (−) BI97D6 and (+) BI97D6 were evaluated side by side with compound (±) BI97D6 in a prostate cancer xenograft using PC-3 cell line in which Mcl-1 and Bcl-X L were overexpressed (Placzek et al., 2010). A quantity of 1 × 10 6 PC-3 cells were injected subcutaneously in the left flank of male athymic nude mice, and the tumors were allowed to grow to an average size of ≈75 mm 3 prior to initiation of therapy. Compounds (±) BI97D6, (−) BI97D6, and (+) BI97D6 were administrated (i.p.) every 2 days (total of nine injections) at two doses of 3 and 5 mg/kg (5 and 8 μmol/kg), respectively. All tested compounds, (±) BI97D6, (−) BI97D6, and (+) BI97D6, displayed a marked inhibitory effect of tumor size compared with the control group, particularly at the dose of 5 mg/kg ( Figure 3D; Figure A2D in Appendix). In fact, the most potent compound (−) BI97D6 induced near complete inhibition of tumor growth at the dose of 5 mg/kg compared with the control group ( Figure 3D). Since (−) BI97D6 displayed better activities against Bcl-X L and Mcl-1 in FP and PC-3 cell assays compared to its (+) isomer, the (−) atropisomer BI97D6 displayed better PC-3 tumor growth inhibitory effect compared to its (+) atropisomer at both doses of 3 and 5 mg/kg ( Figure 3D). All mice tolerated the treatment well with no apparent signs of toxicity in this in vivo assay. The most potent (−) BI97D6 is currently undergoing pharmacokinetic and toxicity studies to decide whether it has scientific merit for further development as a new apoptosis-based cancer drug.

CONCLUSIONS
In summary, (−) and (+) atropisomers of compounds BI97D6 and ApoG2 were synthesized and evaluated in a variety of in vitro and in vivo assays. The optically pure compound (−) BI97D6 was found to bind to Bcl-X L , Bcl-2, Mcl-1, and Bfl-1 with IC 50 values of 76 ± 5, 31 ± 2, 25 ± 8, and 122 ± 28 nM, respectively in FP assays. The compound also potently inhibited growth in culture of the PC-3 human prostate cancer and H23 human lung cancer cell lines with EC 50 values of 0.22 ± 0.08 and 0.14 ± 0.02 μM, respectively. Compound (−) BI97D6 effectively induced apoptosis of the BP3 human lymphoma cell line in a dose-dependent manner and potently killed MEF/WT cell while it showed little cytotoxicity against MEF bax −/ − /bak −/ − cells, suggesting that it killed cancers cells predominantly via the intended mechanism. Finally, (−) BI97D6 demonstrated favorable in vitro ADME properties and superior in vivo efficacy in transgenic mice, in which Bcl-2 is overexpressed in splenic B-cells and further demonstrated greater single-agent antitumor efficacy in a prostate cancer mouse xenograft model at the dose of 5 mg/kg. Given the critical roles of anti-apoptotic Bcl-2 family proteins in tumorigenesis, chemoresistance, and the potent inhibitory effect of (−) BI97D6 against anti-apoptotic Bcl-2 family proteins, we conclude that the reported (−) BI97D6 represent a viable drug candidate for the development of novel apoptosis-based cancer therapies. www.frontiersin.org