Combretastatin-A4 (CA-4) is a natural derivative of the African willow tree Combretum caffrum. CA-4 is one of the most potent antimitotic components of natural origin, but it is, however, intrinsically unstable. A novel series of CA-4 analogs incorporating a 3,4-diaryl-2-azetidinone (β-lactam) ring were designed and synthesized with the objective to prevent cis -trans isomerization and improve the intrinsic stability without altering the biological activity of CA-4. Evaluation of selected β-lactam CA-4 analogs demonstrated potent antitubulin, antiproliferative, and antimitotic effects in human leukemia cells. A lead β-lactam analog, CA-432, displayed comparable antiproliferative activities with CA-4. CA-432 induced rapid apoptosis in HL-60 acute myeloid leukemia cells, which was accompanied by depolymerization of the microtubular network, poly(ADP-ribose) polymerase cleavage, caspase-3 activation, and Bcl-2 cleavage. A prolonged G2M cell cycle arrest accompanied by a sustained phosphorylation of mitotic spindle checkpoint protein, BubR1, and the antiapoptotic proteins Bcl-2 and Bcl-xL preceded apoptotic events in K562 chronic myeloid leukemia (CML) cells. Molecular docking studies in conjunction with comprehensive cell line data rule out CA-4 and β-lactam derivatives as P-glycoprotein substrates. Furthermore, both CA-4 and CA-432 induced significantly more apoptosis compared with imatinib mesylate in ex vivo samples from patients with CML, including those positive for the T315I mutation displaying resistance to imatinib mesylate and dasatinib. In summary, synthetic intrinsically stable analogs of CA-4 that display significant clinical potential as antileukemic agents have been designed and synthesized.
Combretastatin-A4 (CA-4) is a naturally occurring cis -stilbene originally isolated from the African willow tree Combretum caffrum (Pettit et al., 1987). CA-4 is one of the most potent antimitotic and antiangiogenic agents of natural origin. The water-soluble prodrug of CA-4, CA-4P, is currently under development as a vascular targeting agent. Vascular targeting agents are defined as drugs that induce rapid and selective vascular dysfunction in tumors (Young and Chaplin, 2004). CA-4P successfully reduced tumor blood supply and consequently retarded tumor growth in a wide spectrum of preclinical tumor models and in clinical trials (Griggs et al., 2001). CA-4 is structurally and functionally similar to microtubule-targeting agent (MTA) colchicine. Both compounds contain a trimethoxybenzene group, a moiety common to all colchicine-like agents (McGown and Fox, 1989). The combretastatins inhibit the assembly of tubulin by interacting with tubulin at or near the colchicine binding site. Colchicine is one of the first MTAs identified; however, a high toxicity profile prevented clinical progression. Unlike colchicine, CA-4P can inhibit tumor blood flow at 10% of the maximum tolerated dose (Dark et al., 1997). However, recent studies have highlighted the intrinsic instability of the combretastatins caused by isomerization into a more thermodynamically stable and significantly inactive trans-isomer. Cis–trans isomerization readily occurs in heat, light, and protic media and thus limits the potential therapeutic index of this class of MTAs. To circumvent the problem of cis–trans isomerization a substantial range of CA-4 analogs have been designed and synthesized with the objective to prevent cis–trans isomerization and improve the intrinsic stability and the therapeutic index of CA-4. A review by Hsieh et al., (2005) collates this vast array of chemistry focusing on the stabilization of the two aryl rings of CA-4 using one to three atom bridgeheads. CA-4 analogs identified to date include: fluorinated (Alloatti et al., 2008), macrocyclic (Mateo et al., 2007), naphthalene (Alvarez et al., 2007), and imidazole-based (Bellina et al., 2006) analogs. Our group recently designed a novel series of synthetic analogs of CA-4 based on the β-lactam scaffold structure. We have demonstrated that β-lactam scaffolds incorporating appropriately substituted aryl rings at N-1 and C-4 displayed potent antiproliferative effects in human breast cancer cells and inhibited the assembly of purified tubulin in vitro (M. Carr et al., submitted; N. M. O'Boyle et al., submitted). Strategic insertion of the 3,4-diaryl-2-azetidinone (β-lactam) ring prevents the cis–trans isomerization observed with CA-4, creating a series of stable cis-restricted analogs. Plasma stability studies for a selected β-lactam compound demonstrated 92% drug stability after 20 h, leading us to envision a favorable in vivo performance.
Chronic myeloid leukemia (CML) is a malignancy of the hematopoietic stem cell characterized by the expression of the Philadelphia chromosome and its corresponding oncogene, Bcr-Abl. The Bcr-Abl protein is central to the pathogenesis of the disease, a trait that prompted the strategic development of Bcr-Abl targeting therapies. The overall survival and treatment of patients with CML was revolutionized by the introduction of the tryrosine kinase inhibitor imatinib myselate. However, despite the profound clinical success of imatinib mesylate in the treatment of CML, the emergence of imatinib mesylate-resistant phenotypes has rekindled interest in Bcr-Abl independent therapies. In this article, we extend our study of the β-lactam CA-4 analogs to cells of hematological origin, including those displaying resistance to imatinib mesylate and MTAs currently used within the clinic and ex vivo patient samples. We confirm that the promising apoptotic potential of CA-4 and the equipotent CA-432 analog observed in cell lines can be extended to ex vivo samples from patients with CML, including those harboring the T315I mutation displaying resistance to the tyrosine kinase inhibitors imatinib mesylate and dasatinib. The T315I is the most common Bcr-Abl mutation and confers resistance to imatinib by altering the conformation of Bcr-Abl and subsequently preventing the formation of the critical hydrogen bond between imatinib and Bcr-Abl. A comprehensive analysis of the effects of CA-4 and the selected potent β-lactam analog CA-432 on cellular microtubule structure, cell cycle, and apoptosis was determined. We also describe a novel insight into the signaling pathway of combretastatins with particular emphasis on mitotic checking and post-translational modifications to the antiapoptotic proteins Bcl-2 and Bcl-xL.
Materials and Methods
Drug generic names are given unless stated otherwise. CA-4 and β-lactam analogs were synthesized as described by M. Carr et al. (submitted) and N. M. O'Boyle et al. (submitted). CA-4 and analogs were dissolved in ethanol as 10 mM solutions and stored in the dark at −20°C. Vincristine, Adriamycin (generic name; doxorubicin), and paclitaxel were purchased from Sigma Chemical Co. (Poole, Dorset, UK), prepared, and stored according to the manufacturer's instructions. SN-38 (7-ethyl-10-hydroxycamptothecin; generic name of prodrug irinotecan) was a gift from Prof. Patrick Johnston (Queens University, Belfast, Northern Ireland). Unless stated otherwise all general materials were purchased from Sigma Chemical Co. Imatinib mesylate was kindly provided by Witte-Maria Weber, Novartis (Basel, Switzerland) and prepared as a 10 mM stock solution in sterile DMSO.
K562 and HL-60 cells were originally obtained from the European Collection of Cell Cultures (Salisbury, UK). The K562 cells were derived from a patient in the blast crisis stage of CML. HL-60 cells were derived from a patient with acute promyelocytic leukemia. HL-60-BCRP (breast cancer-resistant protein) and HL-60-PGP (p-glycoprotein) cells were generously provided by Prof. Balazs Sarkadi, Hungarian Academy of Sciences, Budapest, Hungary. PGP and BCRP are drug efflux transporters of the ATP binding cassette family of proteins (Staud and Pavek, 2005). A2780-parental and A2780-PGP cells were a kind gift from Prof. Robert Brown, Beatson Institute of Cancer Research, Glasgow, Scotland. Baf/3 murine pro-B cell lines transfected with either native or TS15I mutant Bcr-Abl were provided by Prof. Michael W Deininger, Oregon Health and Science University, Portland, OR. Peripheral blood mononuclear cells (PBMCs) were isolated from heparinized peripheral blood of CML patients by Lymphoprep (Axis-Shield, Oslo, Norway) density gradient centrifugation. All cells were cultured in RPMI 1640 Glutamax medium supplemented with 10% fetal bovine serum media, 100 units/ml penicillin, and 100 μg/ml streptomycin. Cells were maintained at 37°C in 5% CO2 in a humidified incubator. Cell culture materials were supplied by Invitrogen (Carlsbad, CA).
Alamar Blue Cell Viability Assay.
The cytotoxic effects of CA-4 and selected β-lactam analogs on leukemia cells were determined by using the Alamar Blue assay (Invitrogen). The reduction of Alamar Blue is proportional to the number of viable cells. Cells (200 μl) were plated in triplicate in 96-well plates (100,000/ml, K562; 300,000/ml, HL-60; 50,000/ml, A-2780; 150,000/ml, Baf/3). A-2780 cells were plated 24 h before treatment. Suspension cells were plated in the log phase of growth and treated immediately. The cells were then treated with either medium alone, vehicle [1% ethanol (v/v) or 0.1% DMSO (v/v)], or a range of drug concentrations (0.001–10 μM). After 72 h, Alamar Blue was added to each well (10% of final volume), and fluorescence was read by using a 96-well fluorimeter with excitation at 530 nm and emission of 590 nm. The blank solution consisted of medium and Alamar Blue and was used to calibrate the spectrophotometer to zero absorbance. The relative cell viability (%) related to control wells was calculated by [A]test/ [A]control × 100, where [A]test is the absorbance of the drug-treated cells and [A]control is the absorbance of the vehicle-control treated cells. Dose-response curves were plotted, and IC50 values (concentration of drug resulting in 50% reduction in cell survival) were obtained by using the commercial software package Prism (GraphPad Software, Inc., San Diego, CA). Experiments were performed in triplicate on at least three separate occasions.
Immunofluorescence and Confocal Microscopy.
After treatment, cells were cytocentrifuged, air-dried, and fixed in methanol for 30 min at −20°C. The number of MTA-treated cells cytocentrifuged was 4-fold higher than control cells to allow for the reduction in cell number caused by cell death. After washes in phosphate-buffered saline and 0.1% Triton X-100 (PBST), cells were blocked in 5% bovine serum albumin diluted in PBST (blocking buffer). Cells were then incubated with mouse antitubulin (DM1A; 1:20; Merck Biosciences, Nottingham, UK) for 1 h at room temperature. After washes in PBST cells were incubated with secondary antibodies (FITC anti-mouse; Jackson ImmunoResearch, Suffolk, UK) for 1 h at room temperature. After washes in PBST, the cells were stained with propidium iodide (PI) at 0.2 μg/ml in PBS for 2 min, mounted in 4% propyllgallate in PBS/glycerol. Confocal images were captured by using an Olympus (Tokyo, Japan) 1 × 81 microscope coupled with Olympus FluoView version 1.5 software. All images in each experiment were collected on the same day by using identical parameters.
Tubulin Polymerization Assay.
Tubulin depolymerization was quantified by using a modified version of a method originally documented by Minotti et al., (1991). In brief, after treatment cells were pelleted, washed in PBS, and harvested into microtubule preserving buffer [0.1 M Pipes, pH 6.9, 2 M glycerol, 5 mM MgCl2, 2 mM EGTA, 0.5% Triton X-100, and EDTA-free protease inhibitor cocktail tablets (Roche Diagnostics, Mannheim, Germany)]supplemented with 4 μM paclitaxel. Paclitaxel is required to maintain the stability of assembled microtubules during the isolation procedure. The soluble fraction (containing depolymerized tubulin) was separated from the insoluble fraction (containing polymerized tubulin) by centrifugation at 16,000g for 45 min at 4°C. The insoluble fraction was then resuspended in 62.5 mM Tris-Hcl, pH 6.8, 6 M urea, 2% SDS, 10% glycerol, and 0.00125% bromphenol blue and briefly sonicated. Equal amounts of each sample were analyzed by Western blotting using antitubulin mAb (DM1A; 1/1000 dilution).
Flow Cytometric Cell Cycle Analysis.
The flow cytometric evaluation of cellular DNA content was performed as described previously (Greene et al., 2008). In brief, after treatment cells were fixed in 70% ethanol, treated with RNase A, and stained with PI. The PI fluorescence was measured on a linear scale by using a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA). The amount of PI fluorescence is directly proportional to the amount of DNA present in each cell. Data collection was gated to exclude cell debris and cell aggregates. At least 10,000 cells were analyzed per sample. All data were recorded and analyzed with CellQuest software (BD Biosciences).
Annexin V Staining.
The percentage of apoptosis in ex vivo CML cells was determined by annexin V staining. Peripheral blood (10 ml) was collected with informed consent from newly diagnosed treatment naive (n = 4) or blast crisis imatinib-resistant patients with CML (n = 2) in EDTA-anticoagulant tubes. Ethical approval for all work carried out on ex vivo samples from patients with CML was obtained from St. James Hospital and Adelaide and Meath Hospital incorporating guidelines from the National Children's Hospital Ethics Committee, Dublin. PBMCs (1 × 106) were treated with vehicle or 250 nM CA-4, CA-432, or imatinib mesylate for 72 h. Cells were collected by centrifugation at 400g for 5 min and resuspended in anti-CD-45 diluted 1:50 in RPMI medium 1640. After a 10-min incubation in the dark at room temperature, cells were centrifuged and washed in Annexin binding buffer (Biosource, Nivelles, Belgium). Cells were again centrifuged and resuspended in Annexin-V-FITC (IQ Products, Groningen, The Netherlands) diluted in Annexin binding buffer (1:50). Samples were next incubated in the dark on ice for 15 min. Annexin binding buffer (1 ml) was added to each sample. Samples were collected by centrifugation and resuspended in 0.5 ml of Annexin binding buffer. Cells were read immediately by flow cytometry and analyzed by CellQuest software. CML cells were selected and gated based on their low to medium side scatter and low CD45 expression. The T315I mutation was detected by using an allele-specific screen that specifically detected the base mutation in the cDNA (Willis et al., 2005). The mutation was confirmed by using direct sequencing (Soverini et al., 2004).
Whole-cell lysates of treated cells were prepared as described previously (Greene et al., 2008). Equal amounts of protein were resolved by SDS-PAGE, transferred to polyvinylidene difluoride membrane, and blocked for 1 h at room temperature in Tris-buffered saline, pH 7.6/0.05% Tween 20 containing 5% (w/v) dried milk (blocking buffer). Membranes were then probed with anti-Bub3 (BD Transduction Laboratories, Cowley, UK), anti-BubR1 (BD Transduction Laboratories), anti-PARP, anti-caspase-3, anti-Bcl-2, anti-Bcl-xL, and antiactin (all purchased from Merck Biosciences) followed by horseradish peroxidase-conjugated secondary antibody and autoradiography with enhanced chemiluminescence (GE Healthcare, Little Chalfont, Buckinghamshire, UK).
The statistical analysis of experimental data was performed by using Student's paired t test. Results are presented as mean ± S.E.M. A value of P < 0.05 was considered to be significant.
Preliminary Molecular Docking Study.
The probability of PGP interactions with CA-4 and selected analogs was assessed by using FRED version 2.2.3 (OpenEye Scientific Software, Santa Fe, NM) in conjunction with Chemgauss2. For ligand preparation, all compounds were inputted by using MOE 2009.10 (Molecular Operating Environment, Chemical Computing Group, Montreal, Canada), and database processing was achieved by using Pipeline Pilot (Accelrys, San Diego, CA). A detailed description of experimental design has been published by Nathwani et al., (2010). The current database consisted of 100 PGP substrates and 76 nonsubstrates as described by Penzotti et al., (2002) supplemented with a series of β-lactam and heterocyclic combretastatin analogs.
Analysis of the Effects of CA-4 and a Novel Series of β-Lactam Analogs on the Cell Viability of Leukemia Cells of Different Hematological Origin.
Given the clinical success of other MTAs in the treatment of a broad spectrum of leukemias, the antileukemic effects of CA-4 and representative β-lactam analogs were evaluated in multidrug-resistant K562 cells (chronic myeloid leukemia) and rapidly proliferating HL-60 cells (acute myeloid leukemia). The structures of combretastatin and selected active analogs (CA-432, CA-104, CA-162, CA-180, and CA-165) and two relatively inactive analogs (CA-153 and CA-159) are shown in Fig. 1. All compounds feature the 3,4,5-trimethoxyphenyl substituent at N-1, which is characteristic of the CA-4 structure (ring A) and of colchicine. Dose-response curves are shown in Fig. 2 . β-Lactams CA-432 and CA-104 were among the most active, displaying antiproliferative activity in the low nanomolar range in both cell types (Fig. 2). CA-432 and CA-104 contain the 3-hydroxy-4-methoxyphenyl substitution pattern present in CA-4 (ring B). In addition, CA-432 contains the aryl substituent at C3, whereas CA-104 is unsubstituted at C3. CA-432 is between 4- and 6-fold more active than CA-104, suggesting that the C3 phenyl group contributes to the potent activity of CA-432. Compounds CA-162 and CA-180 both feature the 4-methoxyphenyl substituent at C-4 (ring B), but lack the characteristic 3-OH group, and were also active in the nanomolar range. CA-165 containing the C-3 aryl group with 2,3,4-trimethoxy substitution on ring B maintained good antiproliferative activity (Fig. 2). β-Lactams CA-153 and CA-159 contain a trimethoxy substitution on both rings A and B, which adversely affected the antiproliferative activity demonstrating activity in the supramicromolar range in breast cancer (M. Carr et al., submitted) and leukemia cells (Fig. 2). In summary, as demonstrated previously, CA-432 was identified as the most potent antiproliferative β-lactam analog with activity in the low nanomolar range that is comparable with CA-4. As anticipated, CA-153 and CA-159 displayed significantly less antiproliferative activity compared with other CA-4 analogs, demonstrating the structural limitations of multiple substitutions on the C-4 aryl ring.
Effects of CA-4 and Selected Analogs on the Cell Cycle and Apoptosis.
It is well established that exposure to MTAs leads to malformed mitotic spindles, mitotic arrest, and apoptosis. Hence, we next examined the effects of CA-4 and the β-lactam analogs on the cell cycle and apoptosis by flow cytometric analysis of propidium iodide-stained cells. The percentage of apoptotic cells was determined by the quantification of the pre-G1 peak-containing cells with hypodiploid nuclei. A mitotic block was represented by an increase in the percentage of cells in the G2M phase of the cell cycle. All active CA-4 analogs induced a dose-dependent mitotic block in K562 cells at 16 h with IC50 values within the nanomolar range (Fig. 3 A). Again, CA-432 was identified as the most active analog in terms of antimitotic properties with an IC50 value in the low nanomolar range and comparable with CA-4. As anticipated, the less active analogs CA-153 and CA-159 had no effect on the cell cycle up to 10 μM. Hence the effects of CA-4 and the most active β-lactam analog, CA-432, on the cell cycle profile of HL-60 cells were determined. In contrast to K562 cells, the combretastatins exhibited a dual effect on the cell cycle profile of HL-60 cells at 16 h. In addition to a marked increase in the percentage of G2M-arrested cells (Fig. 3B), the combretastatins induced a significant amount of apoptosis (subdiploid peak) in HL-60 cells (Fig. 3C). Given the ease of synthesis and the increased stability of CA-432 over the parent compound CA-4, a time course in K562 and HL-60 cells was carried out to investigate further the effect of the representative combetastatin, CA-432, on the cell cycle and apoptosis over time. In both cell lines, a CA-432-induced mitotic block was observed as early as 4 h, with the percentage of G2M-arrested cells increasing by 8 h. The 16-h time point was identified as the point of divergence in terms of apoptotic response to CA-432. Specifically, at this time point the percentage of CA-432 G2M-arrested HL-60 cells declined as the percentage of apoptotic cells increased. In contrast, the majority of CA-432-treated K562 cells remained arrested in the G2M phase up to 48 h (Fig. 3D). Taken together, these findings indicate a rapid induction of CA-432-induced apoptosis in HL-60 cells with a delayed apoptotic response observed in K562 cells exposed to the microtubule targeting agent CA-432.
CA-4 and Its Novel cis-Restricted β-Lactam Analog CA-432 Disrupt the Microtubule Network of Human Leukemia Cells.
The effects of CA-4 and selected active (CA-432) and inactive (CA-153) analogs on the microtubule network of human K562 and HL-60 leukemia cells were examined by sedimentation assays and confocal microscopy. A cellular tubulin polymerization assay was used to confirm a shift in the natural equilibrium between polymerized (insoluble) and unpolymerized (soluble) tubulin. Sedimentation assays followed by Western blotting demonstrated an early decrease (after 4 h) in the ratio of polymerized (insoluble) to unpolymerized (soluble) tubulin in cells exposed to CA-432 and CA-4 (Fig. 4 A). As expected, treatment with the tubulin polymerizer paclitaxel caused a marked increase in polymerized tubulin (Fig. 4A). Untreated cells demonstrated an even distribution between the polymerized and unpolymerized forms of cellular tubulin (Fig. 4A). Confocal analysis of the normal microtubule network demonstrated spoke-like structures radiating from either the center or the periphery of control cells (Fig. 4, B and C). However, early morphological changes in the microtubular structure in response to the combrestatatins were difficult to detect (data not shown). This may be explained by the compacted microtubular network of suspension cells compared with the dense extended cytoskeletal structures attributed by adherent cells. Late changes in the organization of tubulin in HL-60 cells were reported previously for paclitaxel (Grzanka et al., 2005). In agreement with the aforementioned study, late changes in the microtubule organization of both cell lines in response to combretastatins are shown in Fig. 4, B and C. A loss of organized microtubule structure was observed in MTA (CA-4, CA-432, and paclitaxel)-treated K562 cells. Tubulin staining was diffuse and disorganized with punctate staining in parts after prolonged exposure to combretastatins (CA-4 and CA-432). In paclitaxel-treated cells tubulin formed tightly organized ring structures surrounding multinucleated cells. In addition, microtubule bundles were observed in K562 cells exposed to paclitaxel. In contrast, tubulin staining seemed to associate with apoptotic bodies in HL-60 cells after a prolonged exposure to MTAs. Rigid polymerized microtubule structures were also observed in paclitaxel-treated HL-60 cells. The combretastatin analog CA-153 had no effect on the microtubule network of either K562 or HL-60 cells, suggesting structure-specific tubulin activity (Fig. 4, B and C). Collectively, these results demonstrate that CA-4 and its analog CA-432 depolymerize the microtubule network of leukemia cells with the morphological appearance of the microtubule network depending on the extent of apoptotic changes within the cells.
Effects of CA-432 on Selected Regulators of the Mitotic Spindle Checkpoint, Members of the Bcl2 Family, and Apoptotic Markers.
Previous studies have demonstrated phosphorylation of mitotic spindle checkpoint proteins and members of the Bcl2 family during mitosis in response to microtubule disruption (Srivastava et al., 1998; Greene et al., 2008). Given that CA-432 was an effective inducer of mitotic arrest in K562 cells we next examined the effect of CA-432 on the expression of two mitotic spindle checkpoint proteins (BubR1 and Bub3) and two members of the Bcl2 family (Bcl2 and Bcl-xL). As shown in Fig. 5 A, CA-432 caused the phosphorylation of BubR1, Bcl2, and Bcl-xL with no change observed in the expression levels or phosphorylation status of Bub3. A decrease in the expression levels of the antiapoptotic proteins Bcl-2 and Bcl-xL was also observed in response to CA-432 (Fig. 5, A and B). CA-432 induced a dose-responsive effect on the phosphorylation of Bcl2 and Bcl-xL that coincided with the extent of G2M cell cycle arrest. On the other hand, a dose-responsive change in BubR1 phosphorylation was not observed, perhaps indicating that BubR1 is more sensitive to changes in tubulin polymerization dynamics.
As mentioned, flow cytometric analysis of CA-432-treated K562 and HL-60 cells demonstrated differential apoptotic responses. HL-60 cells undergo rapid apoptosis, whereas K562 cells display a more rigid mitotic response with late apoptosis. Hence, we next sought to confirm the apoptotic population of cells by Western blot analyses of two biochemical markers of apoptosis (PARP cleavage and caspase-3 cleavage) over time. The antibody for caspase-3 used in this study detects the full-length procaspase-3. It may be inferred that the disappearance of full-length caspase-3 is indicative of caspase-3 cleavage and subsequent activation. The onset of CA-432-induced apoptosis directly correlated with PARP cleavage and the disappearance of procaspase 3 in both cell lines.
It is noteworthy that a differential effect on Bcl2 expression levels in response to a prolonged exposure to CA-432 was also observed in both cell lines (Fig. 5B). In K562 cells the majority of Bcl2 was hyperphosphorylated up to 48 h and associated with sustained mitotic block. Accordingly, in HL-60 cells the phosphorylation status of Bcl2 also correlated with the extent of CA-432-induced mitotic block. In addition, Bcl2 cleavage correlated with the onset of apoptosis in HL-60 cells. Cleavage of Bcl2 has been described as a proapoptotic event (Kirsch et al., 1999). In contrast, cleavage of Bcl2 was not observed in K562 cells. The effect of CA-432 on a second member of the Bcl2 family, Bcl-xL, was also observed. CA-432 induced the phosphorylation of Bcl-xL, which associated with the extent of G2M cell cycle arrest. Levels of phosphorylated Bcl-xL decreased as the percentage of cells blocked in mitosis declined. Likewise, the phosphorylation of the mitotic kinase BubR1 decreased as the percentage of cells in G2M declined. No change in the levels of the mitotic checkpoint protein, Bub3, was observed in K562 cells exposed to CA-432. BubR1 and Bub3 are expressed at low to undetectable levels in HL-60 cells (Greene et al., 2008), hence the effects of the mitotic checkpoint proteins were not assessed in CA-432-treated HL-60 cells. Overall, these findings support data obtained from flow cytometry suggesting a rapid induction of apoptosis in HL-60 cells and a delayed apoptotic induction preceded by a prolonged mitotic arrest in K562 cells after exposure to the combretastatin analog CA-432.
Molecular Modeling Studies to Determine the Probability of PGP Interactions with CA-4 and Selected cis-Restricted Analogs.
The apoptotic efficacy of many chemotherapeutic drugs can be severely hindered by cellular efflux by the multidrug-resistant protein PGP. Levels of PGP have been shown to correlate with paclitaxel resistance in vitro (Mechetner et al., 1998). Hence, we next carried out a computational docking experiment to identify possible PGP interactions with CA-4 analogs incorporating a β-lactam scaffold. The β-lactam ring provides a similar angle between two phenyl rings as the cis confirmation of CA-4. Conformers were generated by using molecular modeling software and docked into the QZ59-RRR substrate binding of PGP. Poses generated from the docking step were scored. The scoring function was validated with known PGP substrates, including paclitaxel, adriamycin, and vincristine, which ranked at positions 13, 10, and 32, respectively. The cis conformation of CA-4 ranked at 74, whereas its trans counterpart came in at position 131. All cis-restricted β-lactam analogs shown in Fig. 1 ranked higher than the cis conformation of CA-4 with values ranging from 82 to 142. This finding indicates that CA-4 is not a PGP substrate and suggests that substitution of the ethylene bridge with a β-lactam ring is unlikely to influence interactions with the multidrug-resistant protein PGP.
CA-4 and CA-432 Display Potent Cytotoxic Effects on Multidrug-Resistant Cells.
Given that the probable exclusion of the aforementioned β-lactam analogs as PGP substrates is limited to approximately 70% accuracy, the antiproliferative effects of CA-4 and a selected β-lactam analog, CA-432, were next evaluated in multidrug-resistant cells. Three drug-resistant cell lines and respective parental cell lines were assessed. Specifically, two cell lines of different neoplastic origin [HL-60; acute myeloid leukemia (AML), A2780; ovarian carcinoma] overexpressing the PGP and another (HL-60) overexpressing a structurally distinct multidrug-resistant protein, BCRP, were tested. Cells were exposed to CA-4, CA-432, and selected drugs required to confirm drug resistance. Western blot analysis confirmed the overexpression of PGP and BCRP in respective cell lines (Fig. 6 A). As shown in Table 1, the calculated resistant factors demonstrate that neither CA-4 nor CA-432 display cross-resistance with other microtubule-targeting agents, paclitaxel and vincristine, or adriamycin and SN-38. We next examined the ability of CA-4 and CA-432 to induce apoptosis in a selected resistant cell line, HL-60-PGP. As shown in Fig. 6, B and C, both CA-4 and CA-432 induced apoptosis with equal potency in both HL-60 parental and HL-60-PGP cells. The percentage of apoptosis was calculated by quantification of the pre-G1 peak and confirmed by PARP cleavage. Treatment with CA-4 and CA-432 also resulted in cleavage of the Bcl2 antiapoptotic protein, an event frequently associated with apoptosis (Fig. 6C). Taken together, these results complement PGP molecular docking predictions and confirm that, unlike other MTAs (paclitaxel and vincristine), CA-4 and CA-432 are poor substrates for PGP.
CA-4 and the β-Lactam Analog CA-432 Are Potent Inhibitors of CML Cells Expressing the T315I Mutation.
Imatinib mesylate is the first-line treatment for Bcr-Abl-positive CML. Mutations at critical points in the Bcr-Abl gene are responsible the majority of imatinib resistance in patients with CML. Therefore, we next examined the antiproliferative activity of CA-4 and CA-432 in Baf/3 cells expressing the most abundant Bcr-Abl mutation, T315I. As shown in Table 1, both CA-4 and CA-432 were equipotent in Baf/3 cells expressing native and T315I-mutated Bcr-Abl.
CA-4 and CA-432 Induce Apoptosis in Ex Vivo Imatinib Mesylate Naive and Resistant BCR-ABL-Positive CML Cells.
This study is the first preclinical evaluation of combretastatins in the treatment of chemotherapy naive and patients with acquired imatinib mesylate and dasatinib resistance. We compared the apoptotic potency of CA-4 and CA-432 with imatinib mesylate in primary CML cells. As shown in Fig. 7 , both CA-4 and CA-432 induced apoptosis in chemotherapy naive (n = 4; ●) and imatinib mesylate and dasatinib resistant (n = 2; blast phase; ○) primary CML cells. All cells were treated with a clinically achievable concentration of imatinib mesylate (250 nM) (Gambacorti-Passerini et al., 2000). For comparison, both CA-4 and CA-432 were also tested at 250 nM. It is noteworthy that in five of six cases CA-4 and CA-432 were more effective inducers of apoptosis than imatinib mesylate in the patient samples tested, including two patients expressing the T315I mutation. This finding warrants further development of the antivascular combretastatin drugs as antileukemic agents.
Almost a decade ago, tubulin was identified as a molecular target of the oldest recorded treatment for leukemia, arsenic (Li and Broome, 1999). Traditionally, arsenic was used to successfully treat CML (Forkner, 1938). In more recent years, complete remissions were observed in acute promyelocytic leukemia patients treated with arsenic trioxide (Soignet et al., 1998). The vinca alkaloids, a group of tubulin depolymerizers, are routinely used in the treatment of acute lymphoblastic leukemia (Pui and Evans, 1998). In addition, recent studies demonstrated potent selective cytotoxicity of vincristine against ex vivo chronic lymphocytic leukemia (CLL) cells (Vilpo et al., 2000). Likewise, a structurally different tubulin depolymerizer, pyrrolo-1,5-benzoxazepine-15, potently induced apoptosis in poor prognostic subgroups of CLL (McElligott et al., 2009). It is noteworthy that pyrrolo-1,5-benzoxazepine-15 was more effective than fludarabine in inducing apoptosis in CLL cells. Fludarabine, a purine analogue, is currently the front-line agent in CLL therapy. Paclitaxel, a tubulin polymerizer, is effective in the treatment of aggressive forms of non-Hodgkin's lymphoma (Younes et al., 1997). Interaction of leukemic cells with vascular cells has been associated with resistance to chemotherapy (Frankel and Gill, 2004). Hence, the vascular targeting agents, the combretastatins, are the most recent class of MTAs under development as antileukemic agents. Collectively, the clinical success of MTAs from ancient to modern times in the treatment of leukemia has prompted the design and syntheses of analogs based on these lead components, some of which display improved clinical efficacy over the parent compounds. CA-4P is the clinically preferred phosphate analog of CA-4, owing to improved solubility over CA-4. CA-4P induced mitotic catastrophe in human chronic lymphocytic leukemia cells and demonstrated a marked regression of leukemic xenografts (Nabha et al., 2002; Petit et al., 2008). Extensive data obtained from structure–activity relationships demonstrating the critical importance of the cis-stereochemistry of CA-4 in terms of anticancer activity prompted the design and synthesis of a steady stream of cis-restricted analogs offering improved intrinsic stability over CA-4.
In this study, we demonstrate that a recently described series of conformationally cis-restricted CA-4 analogs incorporating a β-lactam ring yield a more favorable antiproliferative values than other strategic approaches aimed at preventing cis-trans isomerization (Mateo et al., 2007). Structure-specific activities were noted among the analogs, providing vital information on the structural basis of combretastatin and a scaffold for the design of superior cis-restricted β-lactam analogs with improved stability and solubility. A selected cis-restricted β-lactam analog, CA-432, depolymerized tubulin, inhibited cell proliferation, and induced mitotic arrest and apoptosis in human CML and AML cells in a similar manner to CA-4.
Cell type-specific responses to CA-4 and its analog, CA-432, were observed in CML K562 and AML HL-60 cells. Cell type-specific apoptotic responses to other MTAs in leukemia cells have been reported (Al-alami et al., 1998). In this study, apoptotic changes in response to a representative combretastatin analog, CA-432, were observed in HL-60 cells as early as 16 h, whereas apoptosis was delayed until 48 h in K562 cells. The apoptotic delay in K562 cells in response to CA-432 was accompanied by a delay in the disappearance of procaspase-3 and cleavage of the caspase substrate PARP. The apparent delay in CA-432-induced apoptosis in K562 cells was associated with a more stringent G2M cell cycle arrest. In contrast, only a transient mitotic block was observed in HL-60 cells in response to CA-432. In K562 cells, a lengthy phosphorylation of the mitotic spindle checkpoint protein, BubR1, and the antiapoptotic proteins, Bcl2 and Bcl-xL, correlated with the CA-432-induced mitotic block and preceded CA-432-induced apoptosis. BubR1 is phosphorylated during normal mitosis and in response to microtubule insult by other MTAs (Li et al., 1999; Greene et al., 2008). We have demonstrated previously that K562 cells express high endogenous levels of BubR1 and HL-60 cells express low to undetectable levels of BubR1 (Greene et al., 2008). High levels of BubR1 were associated with a prolonged mitotic block in response to microtubule insult by a novel series of tubulin depolymerizers, the pyrrolo-1,5-benzoxazepines and nocodazole (Lee et al., 2004; Greene et al., 2008). Furthermore, small interfering RNA experiments targeting BubR1 decreased the length of mitosis and resulted in an impaired mitotic response to MTAs (Meraldi et al., 2004; Sudo et al., 2004). Hence, collectively these data would suggest that phosphorylation of BubR1 may contribute to the sustained mitotic block observed in K562 cells in response to mitotic insult by CA-432 and consequently delaying the onset of apoptosis.
A differential effect on the antiapoptotic protein Bcl2 in response to CA-432 exposure was also observed in leukemia cells. Bcl2 can either inhibit or promote apoptosis depending on the type of post-translational modification. Phosphorylation of Bcl2 augments its antiapoptotic properties, whereas the caspase-3 cleavage product of Bcl2 displays proapoptotic properties (Kirsch et al., 1999; Deng et al., 2004). In response to CA-432 Bcl2 remains extensively phosphorylated until 48 h in K562 cells. This finding supports previously published work demonstrating that Bcl2 is phosphorylated during mitosis in response to mitotic insult (Scatena et al., 1998). Given that phosphorylation of Bcl2 enhances the antiapoptotic properties of Bcl2 it may be suggested that Bcl2 phosphorylation may also contribute to the delay in CA-432-induced apoptosis observed in K562 cells. In contrast, Bcl2 cleavage was evident in HL-60 cells from 16 h, an event associated with the onset of apoptosis (Kirsch et al., 1999). Bcl2 is cleaved by caspase-3 during apoptosis by a variety of apoptotic stimuli. In this study, Bcl2 cleavage was associated with caspase-3 cleavage and activation, suggesting that caspase-3 may also be responsible for Bcl2 cleavage during CA-432-induced apoptosis. No Bcl2 cleavage was observed in K562 cells that displayed limited caspase-3 cleavage. Taken together, these results suggest that the apoptotic outcome in response to CA-432 in leukemia cells may depend on the type of Bcl2 modification.
Contrary to other MTAs currently used within the clinic such as paclitaxel and vincristine, the combretastatins CA-4 and CA-432 demonstrated potent activity in PGP-overexpressing multidrug-resistant leukemia cells. This finding is of clinical significance given that expression of the PGP frequently is associated with clinical resistance to chemotherapy in CML, acute leukemia, and adult T-cell leukemia (Kuwazuru et al., 1990). A similar finding was observed in PGP-overexpressing ovarian cells, arguing against cell type-specific responses of combretastatins and demonstrating a broad activity spectrum. The combretastatins were also active in BCRP-overexpressing HL-60 cells, demonstrating a lack of cross-resistance with SN-38, a topoisomerase I inhibitor. In HL-60-BCRP cells, both CA-4 and CA-432 induced 61 ± 4.6 and 62.4 ± 7.2% apoptosis, respectively, together with cleavage of PARP and the antiapoptotic protein Bcl-2 (data not shown). These data are of clinical significance because expression of BCRP in AML patients is associated with a lower complete response rate. The tyrosine kinase inhibitor matinib mesylate is currently the first-line treatment for CML patients. It is noteworthy that CA-4 and its intrinsically stable analog, CA-432, induced significantly more apoptosis than imatinib mesylate in primary CML cells, including those positive for the T315I mutation and unresponsive to either imatinib mesylate or the second-generation tyrosine kinase inhibitor dasatinib.
Collectively, data presented herein warrant further clinical development of CA-4 and its synthetic analog CA-432 in chemotherapy naive patients with leukemia and chemotherapy-resistant patients. We demonstrate that synthetic derivatives of combretastatin can be designed and synthesized that do not isomerize and display potent antimitotic and antitubulin properties endowed by CA-4. In conclusion, these data provide important structure–activity information that may be exploited in the design of superior combretastatin analogs with improved intrinsic stability and solubility that may advance the clinical development of combretastatin-A4 in the treatment of leukemia and other malignancies.
We thank Dr. Orla Hanrahan (School of Biochemistry and Immunology, Trinity College, Dublin, Ireland) for technical assistance with the confocal microscope. The Molecular Design Group thanks the Chemical Computing Group (MOE software), OpenEye Scientific (OpenEye product suite), and the Trinity Centre for High Performance Computing and Accelrys (Pipeline Pilot) for support.
This work was supported by Health Research Board Ireland [Grant RP/2007/42] and Higher Education Authority Ireland [Grant G03009].
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- acute myeloid leukemia
- chronic myeloid leukemia
- combretastatin-A4 phosphate
- microtubule-targeting agent
- dimethyl sulfoxide
- breast cancer-resistant protein
- peripheral blood mononuclear cell
- phosphate-buffered saline and 0.1% Triton X-100
- fluorescein isothiocyanate
- propidium iodide
- polyacrylamide gel electrophoresis
- poly(ADP-ribose) polymerase
- chronic lymphocytic leukemia
- 1,4-piperazinediethanesulfonic acid
- monoclonal antibody.
- Received May 18, 2010.
- Accepted August 10, 2010.
- Copyright © 2010 by The American Society for Pharmacology and Experimental Therapeutics