Design, synthesis, biological evaluation and cocrystal structures with tubulin of chiral β-lactam bridged combretastatin A-4 analogues as potent antitumor agents
ABSTRACT
A diverse of chiral β-lactam bridged analogues of combretastatin A-4 (CA-4), 3-substituted 1,4-diaryl-2-azetidinones, were asymmetrically synthesized and biologically evaluated, leading to identify a number of potent anti-proliferative compounds represented by 14b and 14c with IC50 values of 0.001-0.021 µM, against four human cancer cell lines (A2780, Hela,SKOV-3 and MDA-MB-231). Structure-activity relationship (SAR) studies on all stereoisomers of 14b and 14c revealed that the absolute configurations of the chiral centers at 3- and 4-position were critically important for the activity and generally a trans configuration between the “A” and “B” rings is optimal. In addition, 14b and 14c displayed less cytotoxicity on normal human oviduct epithelial cells than malignant cells indicating good selectivity in vitro. Further biochemical evaluation and cocrystal structures with tubulin demonstrated that both compounds disrupted tubulin polymerization through interacting at the colchicine-binding site, suppressed angiogenesis in vitro and in vivo, blocked cell cycle progression at mitotic phase and induced cellular apoptosis. The in vivo assays verified that both compounds inhibited xenograft tumor growth in nude mice with acceptable therapeutic window, showing promising potentials for further clinical development.
1.Introduction
Microtubule is an important component of cytoskeleton, which is polymerized by α- and β-subunit. Microtubule plays a role of separating the daughter chromosomes to the opposite poles during the mitosis. The disruption of microtubule will result in the interruption of mitosis and lead to the apoptosis of cell [1]. After Colchicine is the first discovered compound which binds to the dimer and destabilizes microtubule [2], Combretastatin A-4 (CA-4, 1, Figure 1) is a natural product isolated from Combretum caffrum which can bind to the colchicine binding-site, and shows extremely strong depolymerization of microtubule in vitro [3]. Its water soluble phosphate prodrug CA-4P (2), is now in phase III clinical trials [4]. Since the cis-double bond is critically important for the activity of 1 and the spontaneous isomerization from cis-isomer to trans-isomer leads to its chemical instability [5], a large number of conformationally restricted analogues of CA-4 have been reported, in which the cis double bond is replaced by a heterocycle [6]. Among them, β-lactam bridged analogues of CA-4 are the most attractive and a number of substituted 1,4-diaryl-2-azetidinones have been reported with significant suppressions on tubulin polymerization and cellular proliferation [7]. Due to two chiral carbon atoms of azetidin-2-one scaffold and the existence of two pairs of enantiomers, accumulating studies have been focused on the relationship between stereo-chemical structure and anticancer activity. The trans-form of C-3 and C-4 turned out to be favor by the work of Meegan and Coccetti groups [7d,7f]. But the absolute configurations of chiral centers remained unclear until compounds 3/4 were synthesized asymmetrically by our group and firstly identified the favorable orientation of phenyl at 4-position on β-lactam ring which demonstrated critically important for the antiproliferative activity [8]. In this study, we would like to report a series of optically pure β-lactam linked CA-4 analogues to define their SARs systematically, especially the influence of the absolute configurations of C-3 and C-4 on the activities.
2.Results and discussion
A series of chiral 3-methylene-2-azetidinones were synthesized asymmetrically according to literature method [9] (Scheme 1). Racemic MBH adducts 6 were obtained by Morita-Baylis-Hillman reaction of corresponding aromatic aldehydes 5 with ethyl acrylate in the presence of DABCO in 3-14 days, depending on the reactivity of the carbonyl group of aldehydes. After protecting the hydroxyl groups of 6 with acetic anhydride, 7 were submitted to the asymmetric allylic amination using aromatic amines 8 as the nucleophiles under the catalysis of Pd2(dba)3 and 9a, affording the corresponding optically active amination products 10 in high yields (up to 94%) and enantioselectivities (up to 99% ee). Finally 10 can be readily transformed into their corresponding β-lactam derivatives 11a-r in good yields (37-92%) without loss of enantioselectivity by reaction with Sn[N(TMS)2]2 in refluxing toluene.Previous research has indicated clearly that the C-3 substituent is relatively flexible [6b,7a,7c,7d,8,10]. With 3-methylene-2-azetidinones in hand, modifications of 3-methylene group allowed us to further examine the role of 3-substituent as well as the cis/trans/absolute configurations of azetidin-2-ones on their antiproliferative effects. Accordingly, a diversity of derivatives of 3 were designed and synthesized, and these transformations could be readily realized with 3-methylene as latent functional group (Schemes 2-4).Various hydroxyl-protected compounds 12a-h were obtained by acylation or alkylation of 3 with corresponding acyl chlorides or alkyl halides in good yields (Scheme 2). Then olefin cross-metathesis of 3 were performed under the catalysis of Grubbs II to give 3-benzylidene substituted compounds 13a-b, and the newly formed C=C bonds were identified as Z configurations as indicated by NOESY.
The Michael additions of 3 or 11g with corresponding nucleophiles afforded 3-hydroxymethyl or aminomethyl substituted target compounds 14a-j. The trans-diastereoisomer of 4, 14c and 14g could be obtained by bromination of 14b or 14e followed by hydrogenation under the catalysis of Pd/C in the presence of AcONa.In order to further diversify the substitution of C-3, the dihydroxylation of 12e proceeded under the catalysis of K2OsO4·2H2O with 4-methylmorpholine N-oxide (NMO) as the oxidant, yielding 3-hydroxy-3-hydroxymethyl substituted compound 15, whose configuration was determined by NOESY (Scheme 3). The direct catalytic hydrogenolysis or methylation followed by catalytic hydrogenolysis of 15 provided products 16a-b. Oxidation with NaIO4 and subsequent reduction with NaBH4 of 15 afforded cis-(3R,4S)-3-hydroxy substituted analogue 17, whose configuration was determined as indicated by 1H NMR. The alkylation or esterification followed by the debenzylation of 17 produced 3-alkoxy or 3-acyloxy substituted analogues 18a-b. The sulfonylation of 17 with mesyl chloride gave 3-mesyloxy substituted product 19 in 86% yield.Nucleophilic substitution of 19 with tetrabutylammonium bromide (TBAB) under the condition of microwave yielded trans-(3S,4S)-3-bromo substituted product 20 and its configuration was identified as indicated by 1H NMR (Scheme 4). Debenzylation of 20 with H2 catalyzed by Pd/C resulted in the expected product 21 and the further debrominated product 22 in the yields of 62% and 28%, respectively. Nucleophilic substitution of 19 with NaN3 at 120 oC in DMF provided trans-(3S,4S)-3-azido substituted product 23, whose configuration was confirmed by 1H NMR. Then 23 was submitted to azide-alkyne click reaction with propargyl alcohol afforded 3-(1,2,3-triazol-1-yl) substituted analogue 24. Reduction of 23 with SnCl2 produced trans-(3S,4S)-3-amino substituted compound 25. Derivatization of the amino of 25 with acid anhydride or acyl chloride gave the 3-acylamino or 3-sulphonylamino substituted analogues 26a-c.Activities of Anti-proliferation in Vitro. The newly synthesized chiral β-lactam analogues of CA-4 were screened for anti-proliferative activities against four human cancer cell lines A2780, Hela, SKOV-3, and MDA-MB-231 with CA-4 (1) and paclitaxel （PTX）as positive control. As summarized in Table 1, most of target compounds display moderate to potent inhibition on cell growth. Among them, compounds 11g, 11h, 12f, 14b, 14c, 14e, 14g, 21, 22, 24, 26a and 26b exhibited considerable activities with IC50 values less than 100 nM, therein compound 14c exhibited similar activity with the referenced agents CA-4 (1) and PTX. In addition, we also tested the cytotoxicity of 14b and 14c in a normal ovarian epithelial cell line HOSE, and the result shows that both compounds achieve less 50% inhibition ratio at 1 µM, indicating the low toxicity of 14b and 14c against normal cells.
Comparing the data of 4 and 14a-g with 30a-d (Scheme 6), S-configuration of C-4 is critically important for the activity, which showed the same conclusion as the reported literature [8]. 3-Hydroxymethyl substituted products 14b and 14e with trans-(3R,4R) configuration displayed about 4-20 fold more potent than cis-(3S,4R) diastereoisomers 14a and 14d. 3-Methyl substituted compounds 14c and 14g with trans-(3S,4R) configuration also displayed about 4-63 fold more potent than cis-(3R,4R) diastereoisomers 4 and 14f. Analogues (21, 24, 26a-b) with trans-orientation also showed excellent antiproliferative activities. In conclusion, trans-orientation of substituents at 3,4-positions of β-lactam scaffold benefit the activity, which coincide with the results reported by Coccetti and Meegan groups [7d,7f].Many of 3-methylene substituted analogues (11e, 11g, 11h) showed good activities, but 3-(Z)-benzylidene analogues (13a-b) were inactive, indicating that 3-methylene is tolerant but bulky steric hindrance in the same plane is forbidden. The 3,4,5-trimethoxyphenyl turned out to be the best choice for N-1 substitution, which is consistent with CA-4 and previous research results. The analogue with 3,5-dimethoxy substituted N-phenyl ring (11r) exhibited a little loss of activity, whereas 11n and 11p with only one MeO- group on their N-phenyl rings showed considerable loss in potency. The 4-methoxy of the phenyl ring at C-4 was very important for the activity and the lack of 4-methoxy (11a, 11b, 11j) resulted in loss of activity, whereas the 3-hydroxy of 4-phenyl ring can tolerate various arrangements. Removal or replacement of 3-hydroxy of 4-phenyl ring with other bioisosteres, such as F, Cl and NH2, as well as the acylation of 3-hydroxy can maintain the activity (11e, 11f, 11g, 11h, 12f, 12h). In the cases of the alkylated derivatives (12a-e) of 3-hydroxy at 4-phenyl ring, the bulkier steric hindrance of alkyl group the less potent the activity is.cellular microtubule network. CA-4 (1) served as tubulin depolymerized control, whereas PTX was used as tubulin stabilized control. DMSO (0.2%) was used as negative control. Scale bars: 20 µm.Next, the effects of 14b and 14c on the inhibition of cellular tubulin assembly were also determined. After treatment with 14b, 14c, 1 or PTX respectively, the cells were lysed with detergents, and centrifuged.
The pellet (P, polymerized tubulin) and supernatant (S, depolymerized tubulin) fraction were collected and analyzed by western blotting (Figure 2B and S1). Treatment of 1 significantly decreased polymerized microtubules while treatment of PTX increased polymerized tubulin in Hela cells. As expected, cells treated with the two tested compounds showed dramatically reduction of polymerized tubulin.Furthermore, we employed cellular immunofluorescence assay to visualize the effects of β-lactam analogues 14b and 14c on microtubule structure. As shown in Figure 2C, when treated with solvent (DMSO) control, the microtubule network (green) of SKOV-3 cells were slim and hair-like, distributing around nucleus (blue). However, when treated cells with 1, 14b or 14c, the fibrous microtubule structures were disorganized and their density was also significantly reduced. In contrast, cells treated with PTX exhibited a spindle shaped microtubule network and the density of microtubule was significantly increased [11]. Taken together, these results clearly demonstrated that our β-lactam bridged CA-4 analogues strongly inhibit tubulin assembly in both purified protein and cancer cells. Anti-angiogenic Effects in Vitro and in Vivo. Targeting angiogenesis has been one of the strategies for increasing the efficacy of cancer therapeutics. Since many tubulin binding agents, including CA-4, have been reported to interfere with vascular formation [12], 14b and 14c were evaluated for their effects against angiogenesis in vitro and in vivo in this study. In capillary-like tube formation assay, vascular endothelial cells could form capillary-like structures on matrigel [13], an extracellular matrix rich in proangiogenic factors. As shown in Figure 3A, the HUVEC cells only treated with diluent (DMSO) produced a complete network structure. However, both 14b and 14c could effectively alter the tubule-like structures after 12 h incubation. At the concentration of 50 nM, 14c almost completely disrupted the capillary-like tube formation.Furthermore, the in vivo anti-vascular activities of 14b and 14c were also tested using matrigel plug assay [14]. In this assay, matrigel supplemented with human VEGF (100 ng/mL) was mixed with 14b (100 nM), 14c (100 nM) or diluent (DMSO), followed by injection into nude mice subcutaneously. One week later, the growth of capillary blood vessel in the plug was massively induced by VEGF, when matrigel was only mixed with diluent (Figure 3B upper line).
However, the existence of compound 14b or 14c attenuated the VEGF-induced capillary blood vessel in the matrigel plug. The relative level of angiogenesis was also presented by hemoglobin content in the matrigel plug. As shown in Figure 3B lower line, the levels of hemoglobin in matrigel were reduced by 14b and 14c treatment compared with diluent treated control. Collectively, these results demonstrated that 14b and 14c suppressed VEGF-induced angiogenesis in vitro and in vivo. Inhibition of Colony Formation and Cell Cycle Arrest. In the following experiments, we further confirmed the cytotoxic effects of 14b and 14c by colony formation assay. As shown in Figure 4A and 4B, the colonies formed by MDA-MB-231 cells were dose-dependently reduced by the exposure to 14b and 14c.To further characterize the inhibitory effects of 14b and 14c on cell growth, cell cycle progression was analyzed by quantitating DNA content, since chemical agents generally decrease growth and proliferation of cancer cells through altering the regulation of cell cycle [15]. As expected, results showed that, similar to the positive compound 1, both compounds 14b and 14c arrest Hela cells in the G2/M phase (Figure 4C and S2). In the DMSO treated group, only about 24% of cells were distributed in the G2/M phase. However, after being treated with 14c or 1 at the dose of 50 nM, nearly 100% of cells were arrested in G2/M phase. Moreover, immunoblotting showed that 14b and 14c markedly up-regulated the expression of cyclin B1, which is synthesized at G2-phase and reaches maximum level at metaphase. In addition, 14b and 14c treatment also induced hyper-phosphorylation of the mitotic checkpoint kinase BubR1 and mitotic trigger histone H3 in a dose-dependent manner (Figure 4D). These results clearly demonstrate that, consistent with other reported microtubule inhibitors [16], β-lactam analogues 14b and 14c significantly induce cellular mitotic arrest in Hela cells.
Induction of Cellular Apoptosis. Numerous studies have demonstrated that tubulin polymerization inhibitors possess the abilities to induce cellular apoptosis [17]. For this reason, we assessed whether the active analogues 14b and 14c could induce apoptosis of Hela cells by double staining with Annexin V-FITC and propidium iodide (PI). The flow cytometric data (Figure 5A) showed the high activities of 14b and 14c on the induction of cellular apoptosis. As qualified in Figure S3, the total proportions of apoptotic cells, including Annexin V+/PI- (the right lower quadrant representing early apoptosis) and Annexin V+/PI+ (the right upper quadrant representing late apoptosis or necrosis), were 38.7%, 57.5% and 70.7% after treatment with 50, 100 and 200 nM of 14b, and were 52.7%,65.1% and 71.3% after treatment with 50, 100 and 200 nM of 14c, respectively. Exposure to positive control 1 (100 nM) leaded to totally 62.5% cell apoptosis, however, only 2.7% apoptotic cells were detected in solvent (DMSO) treated group. Subsequently, we further examined apoptosis associated proteins by immunoblotting analysis. As indicated in Figure 5B, pro-apoptotic protein Bax and p53 was up-regulated by both 14b and 14c treatment in a dose-dependent manner. In addition, both compounds also dose-dependently promoted cleavage of poly(ADP-ribose) polymerase-1 (PARP-1), a marker of cells undergoing apoptosis. These results implied that the β-lactam analogues exhibited their anti-tumor effects through, at least partly, induction of cellular apoptosis. Anti-tumor Activities in Vivo. Firstly, we performed single-dose acute toxicity assay to assess the safety of compounds 14b and 14c. ICR mice (n=10, half male and half female) were injected intraperitoneally with various dosages of 14b (95, 70, 50, 35 and 25 mg/kg) or 14c (275, 200, 150, 125 and 100 mg/kg), or vehicle control. As summarized in Table 3, none of the mice died in the two days after treatment with 14b, whereas the death of mice mainly occurred in the second day after treatment with 14c. With LD50 values of 136.5 kg/mg, 14c exhibited less toxicity than 14b, whose LD50 value was calculated as 61.5 mg/kg.Next, we evaluated the in vivo anti-tumor effects of compounds 14b and 14c using human ovarian cancer xenograft mice model, which is established by subcutaneous inoculation of A2780 cells in the female Balb/C nude mice.
After the mean value of tumor volumes reached approximately 100 mm3, the mice were randomly divided into 4 groups and administrated with 7.5 mg/kg of 14b, 8 mg/kg of 14c, 10 mg/kg of PTX (as positive control) or vehicle (negative control) by intraperitoneal injection once every two days. As shown in Figure 6A, treatment of both 14b and 14c significantly suppressed the tumor growth, achieving 78.9% and 76.3% reduction in tumor size at the end of the observation period, respectively. When the observation time ended, the mice were sacrificed and the tumors were excised and weighed. The results were illustrated in Figures 6B and 6D, and again confirmed the anti-tumor activities of the two compounds. Slight weight loss was observed in 14b treated animals, but was not statistically significant (Figure 6C). Furthermore, H&E staining of tumor sections showed extensive areas of necrosis or cell death (indicated by black arrow in Figure 6E) in both 14b and 14c treated groups, however, none of abnormal areas were observed in the vehicle treatment group. In addition, we also examined organs of agents treated and vehicle treated mice, including liver, kidney and spleen, using H&E staining (Figure 6F). No detectable abnormalities were observed in the organs examined, indicating the safety of 14b and 14c at the therapeutic dosage. Taken together, considering its LD50 value, it could be proposed that compound 14c possesses a satisfied therapeutic window and is worth for further validation. organic sections. No obvious alternation was observed in the organs examined, including liver, kidney and spleen after treatment with 14b and 14c. ** P<0.01; *** P<0.001.To reveal the specific details of binding, as well as to provide solid basis for further structural optimization, compounds 14b and 14c were soaked into the crystals of a protein complex (T2R-TTL) consisting of two α,β-tubulin heterodimers, the stathmin-like protein RB3, and the tubulin tyrosine ligase. Finally the co-crystal structures of these two compounds with tubulin were determined at 2.56 Å for 14b (PDB code 5XAG) and 2.55 Å for 14c (PDB code 5XAF), respectively (Figure 7). The data collection and refinement statistics was summarized in Table S1. As expected, 14b and 14c occupied the binding site of colchicine with satisfactory electron density. To determine the impact of ligand binding against the overall conformation of tubulin, these two complex structures were subsequently superimposed to the tubulin structure in the absence of ligand (PDB code 5JQG). The RMSD values for the superimposed structures of 14b and 14c were 0.603 Å over 1892 Cα atoms and 0.608 Å over 1917 Cα atoms, respectively, indicating that these two compounds did not affect the global conformation of tubulin. However, the major conformational changes come from two loops of tubulin near the binding site after ligand binding. Similar conformational changes were also observed in the structure of tubulin with colchicine, indicating that 14b and 14c may possess the same inhibitory mechanism as colchicine. In addition, as shown in Figure S4, the conformations of these two loops were not the same in three structures, suggesting that the flexibility of this binding site might be further exploited. Both hydrophobic contacts and hydrogen bonds were observed in the structures of tubulin with 14b and 14c. As shown in Figure 7G, 14b formed a direct hydrogen bond with β-Ala250, as well as water-bridged hydrogen bonds with β-Asn349. However, the main driving force of binding was thought to be shape matching between the compound and surrounding hydrophobic residues, including the residues Leu242, Leu248, Ala250, Leu255, Met259, Val315, Ala316, Ala317, Ile318 of β-tubulin. Notably, as shown in Figure 7I, 14b and 14c were more deeply buried than colchicine in the binding site, which is consistent with their better potency against tubulin polymerization. However, the acetamido group of colchicine was not overlaid in the superimposed structures, providing us a new possible direction for further structural modification.Figure 7. Crystal structures of the tubulin-14b and tubulin-14c complex. Overall view of the complex formed between 14b and tubulin. α-tubulin is shown as green cartoon, and β-tubulin is shown as cyan cartoon. 14b is shown in yellow sphere representation (PDB code 5XAG). (B) Overall view of the complex formed between 14c and tubulin. α-tubulin is shown as green cartoon, and β-tubulin is shown as cyan cartoon. 14c is shown in yellow sphere representation (PDB code 5XAF). (C) 2Fo-Fc electron density (blue) for 14b, and the contour level is set to 1.0 sigma. (D) 2Fo-Fc electron density (blue) for 14c, and the contour level is set to 1.0 sigma. (E) 14b binds to the colchicine site of β-tubulin, which is shown in surface representation. (F) 14c binds to the colchicine site of β-tubulin, which is shown in surface representation. (G) Close-up views of the hydrogen bonds and hydrophobic contacts formed between 14b (yellow sticks) and tubulin. Hydrogen bonds are shown in yellow dashed lines, and corresponding distances were labeled in angstrom. Interacting residues of tubulin are shown in white stick representation, and crystal waters are shown in red sphere representation. (H) Close-up views of the hydrogen bonds and hydrophobic contacts formed between 14c (yellow sticks) and tubulin. Hydrogen bonds are shown in yellow dashed lines, and corresponding distances were labeled in angstrom. Interacting residues of tubulin are shown in white stick representation, and crystal waters are shown in red sphere representation. (I) Superimposition of 14b (cyan stick, PDB code 5XAG) and 14c (green stick, PDB code 5XAF) with colchicine (salmon stick, PDB code 4O2B). 3.Conclusion In summary, a diverse of chiral β-lactam bridged combretastatin A-4 analogues have been synthesized and biologically evaluated. Most of the target compounds displayed moderate to potent anti-proliferative activities against four human cancer cell lines (A2780, Hela, SKOV-3 and MDA-MB-231). The studies on SARs revealed that the absolute configurations of the chiral C-4 were critically important for the activity, more specifically, (S)-configuration for 3-methylene substituted series (11) and the same orientation for other analogues. On this basis, trans-configuration of substituents at 3,4-positions of β-lactam scaffold benefit the antiproliferative activity. Among all synthesized compounds, 14b and 14c turned out to be most potent and were selected for further pharmacological studies after comprehensive consideration. The co-crystal structures of tubulin in complex with 14b and 14c were determined by X-ray crystallography, which showed that they bind to the same site as colchicine with similar binding mode. Further biochemical evaluation demonstrated that both compounds disrupted tubulin assembly in isolated protein level and in cellular level, suppressed angiogenesis in vitro and in vivo, blocked cell cycle progression at mitotic phase and induced cellular apoptosis. Importantly, both compounds inhibited xenografts tumor growth in nude mice with acceptable therapeutic window, showing promising potentials for further clinical development. 4.Experimental section All reagents were commercially available and were used without further purification. Melting points were measured on a SGW X-4 apparatus and uncorrected. Optical rotations were measured at the sodium D line, using a Jasco P-1020 automatic digital polarimeter and a 100 mm cell. 1H and 13C spectra were obtained by a Varian 400 MHz or a Bruker 600 MHz NMR spectrometer at 303 K, using tetramethylsilane as internal standard. MS was measured on Agilent 6120 Quadrupole LC/MS. HRMS determinations for all new compounds were performed on AB SCIWX TRIPLETOF 5600+ or Agilent 6224 TOF LC/MS, respectively. Flash chromatography was carried out using standard silica gel 60 (300−400 mesh) and detected with UV light monitor at 254 nM. The purity of the biological tested compounds was determined by reverse-phase HPLC (SB-CN column (5 Micro, 4.6 mm × 250 mm) using a Waters 1525 Binary HPLC Pump and a Waters 2489 UV/visible Detector). The mobile phase was MeOH/H2O (85:15, v/v) and with a total flow rate of 1 mL/min. The purity was determined by monitoring at 254 nm. The purities of all the activity tested compounds were confirmed to be ≥ 95% by this HPLC Combretastatin A4 analysis.