Tubulin-interactive stilbene derivatives as anticancer agents

Microtubules are dynamic polymers that occur in eukaryotic cells and play important roles in cell division, motility, transport and signaling. They form during the process of polymerization of α- and β-tubulin dimers. Tubulin is a significant and heavily researched molecular target for anticancer drugs. Combretastatins are natural cis-stilbenes that exhibit cytotoxic properties in cultured cancer cells in vitro. Combretastatin A-4 (3′-hydroxy-3,4,4′, 5-tetramethoxy-cis-stilbene; CA-4) is a potent cytotoxic cis-stilbene that binds to β-tubulin at the colchicine-binding site and inhibits tubulin polymerization. The prodrug CA-4 phosphate is currently in clinical trials as a chemotherapeutic agent for cancer treatment. Numerous series of stilbene analogs have been studied in search of potent cytotoxic agents with the requisite tubulin-interactive properties. Microtubule-interfering agents include numerous CA-4 and transresveratrol analogs and other synthetic stilbene derivatives. Importantly, these agents are active in both tumor cells and immature endothelial cells of tumor blood vessels, where they inhibit the process of angiogenesis. Recently, computer-aided virtual screening was used to select potent tubulin-interactive compounds. This review covers the role of stilbene derivatives as a class of antitumor agents that act by targeting microtubule assembly dynamics. Additionally, we present the results of molecular modeling of their binding to specific sites on the α- and β-tubulin heterodimer. This has enabled the elucidation of the mechanism of stilbene cytotoxicity and is useful in the design of novel agents with improved anti-mitotic activity. Tubulin-interactive agents are believed to have the potential to play a significant role in the fight against cancer.


INTRODUCTION
The high incidence of cancer and the high cost of its treatment are important factors driving the search for new and effective chemotherapeutic substances with multi-target activity and no toxicity to normal cells. Microtubules are believed to be one of the more successful targets for cancer chemotherapy. In 1990, the tubulin-interacting agent paclitaxel was introduced as a clinically approved anticancer chemotherapy agent. Since then, the interest in tubulin as a molecular target for anticancer drugs has significantly increased. Over 25% of new drug candidates that entered clinical trials as anticancer agents between 2005 and 2007 exert their effects through interaction with tubulin [1]. Almost all tubulin-interacting agents are natural products or analogs thereof that are well suited to binding to the complex three-dimensional surface of the tubulin polymer. Studies on their interaction with tubulin-binding sites have provided information on the mechanism of anti-mitotic activity of potential drugs used for cancer treatment and allowed scientists to develop new and efficient chemotherapeutics. This review focuses on the anti-carcinogenic properties of natural and modified stilbenes with respect to their interaction with the cellular protein tubulin. The interest in stilbenes as tubulin-interactive agents was inspired by the work of Pettit, who first isolated combretastatins and determined their anti-mitotic activity in the 1980s [2]. Combretastatins are natural cis-stilbenes that are isolated from the bark of the African willow tree (Combretum caffrum). The most prominent representative of this group of compounds, combretastatin A-4 (3'-hydroxy-3,4,4',5-tetramethoxy-cis-stilbene; CA-4; Fig. 3), exerts high antimitotic and anti-angiogenic activities [see 3 for a review]. Presently, phase III clinical trials are underway to test the anticancer effectiveness of a phosphate prodrug of CA-4 (CA-4P, also known as fosbretabulin; trade name Zybrestat). CA-4P is used alone or in combination with traditional chemotherapeutic agents or with radiotherapy [4]. Microtubules, which are composed of -tubulin heterodimers (Fig. 1), are dynamic polymers of eukaryotic cells. A microtubule is a tube constructed from parallel linear polymers (protofilaments) in which the heterodimers are assembled head to tail in a polar fashion ( Fig. 2A). These protofilaments interact laterally to form the cylindrical and helical microtubule wall (Fig. 2B) [9]. Microtubules play important roles in cell division, cell motility, cellular transport, maintaining cell polarity, and cell signaling. Microtubule polymerization is a complex process involving a cooperative assembly of -tubulin dimers followed by GTP hydrolysis [5]. Nogales et al. reported on the crystal structure of -tubulin heterodimers ( Fig. 1) [6]. Each subunit of the heterodimer has one GTP molecule bound to it. The -subunit binds to GTP in an irreversible manner while the GTP bound to the -subunit is exchangeable and hydrolyzes during polymerization [6]. Impaired microtubule function leads to abnormal cell morphology and may pose a challenge to cell survival, leading to apoptosis. There are three distinct phases of microtubule polymerization: nucleation, polymer growth and steady-state equilibrium. In the nucleation phase, an oligomer consisting of 6 to 12 -tubulin dimers is formed. In the polymer growth phase, GTP binds to -tubulin dimers, which leads to microtubule elongation and the formation of the protofilament. This is followed by the steady-state phase, when the addition and dissociation of tubulin subunits is balanced and the net molecular mass of the polymer does not change. Microtubule dynamics have been investigated in vitro using light microscopy of reconstituted microtubules and in live cells using time-lapse video microscopy. Two types of dynamics have been observed: treadmilling and dynamic instability. Treadmilling is a flow of subunits from one end of the microtubule (the minus end) to the other (the plus end); it does not change the average length of the microtubules. Dynamic instability is the process of alternate growth and shortening of microtubule ends. It was first reported by Mitchison and Kirschner [7]. They termed the transition from a growing phase to a shortening phase as a catastrophe and a transition from a shortening phase to a growing phase as a rescue. GTP hydrolysis is necessary for switching between catastrophe and rescue (Fig. 2C, step 4). It has been postulated that GTP hydrolysis changes the conformation of a protofilament from a slightly curved to a more profoundly curved tubulin-GDP structure [8]. Additionally, microtubules are stabilized and destabilized by microtubule-associated proteins (MAPs). Specialized MAPs named plus-andtracking proteins (+TIPs) interact with the microtubule ends and with each other to regulate microtubule dynamics [see 9 for a review].  Microtubule dynamics play an essential role in mitosis [10,11]. The plus ends of the microtubules undergo a series of random elongations and shortenings in the cytoplasm to search and bind the kinetochores of the chromosomes [12]. Metaphase microtubules are highly dynamic. At the onset of anaphase, their dynamics ensure proper chromosomal segregation [13]. The movement of the chromosomes towards the spindle pole is provoked by tension due to the action of motor proteins and shortening of the microtubules [14]. Since the dynamics of tubulin polymerization play important roles in basic cellular functions, microtubules are a target for drugs used in the treatment of several diseases, including cancer [15,16]. Many microtubule-targeted agents are still in clinical trials. Microtubule-interfering agents (MIA), a group that includes known anticancer drugs, exert polymerizing and depolymerizing effects on tubulin assembly. Both effects impair the proper function of microtubules and stop the cell cycle at the G2/M phase. This is accompanied by the activation of the signaling pathways leading to apoptosis. The mechanism combining these two phenomena has not yet been fully explained. The best known depolymerizing agents (inhibitors) include colchicine, vinblastine and vincristine, but other chemotherapeutic drugs are polymerizing agents (enhancers), such as paclitaxel, docetaxel and the polyisoprenyl benzophenones. Studies on the mechanism of anti-mitotic activity of colchicine and other depolymerizing agents have provided evidence that these agents inhibit cell proliferation by perturbing microtubule dynamics [17]. However, anti-mitotic agents show affinity to different binding sites at the -tubulin dimer. Recently, molecular modeling has been applied to explain the mechanism and the mode of ligand binding to a target protein. In the case of tubulin polymerization inhibitors, the determination of the crystallographic structure of the tubulin heterodimer made this possible [6]. Depolymerizing agents bind to the colchicine-binding site located at the interface between the  and  subunits of the tubulin dimer [18] and to the vinblastine-binding site close to the exchangeable GTP site on -tubulin ( Fig. 1) [19]. Additionally, it was demonstrated that unique binding sites may exist. Depolymerizing agents, such as benomyl, estramustine and LY290181 ([2-amino-4-(3-pyridyl)-4H-naphtho(1,2-b)pyran-3-carbonitrile]) do not compete with colchicine and vinblastine for tubulin binding. This suggests that these agents bind to tubulin at distinct sites [16]. Using fluorescence spectroscopy, Chakraborti et al. demonstrated that curcumin, which is known for its anticarcinogenic properties, binds tubulin at a distance 32Å away from the colchicine-binding site. Docking studies suggested that the curcumin-binding site is localized close to the vinblastine-binding site [20]. The vinca alkaloids were the first natural products to enter clinical use. Vinblastine and vincristine have been used in clinical oncology since the 1960s. Vindesine and vinorelbine are semi-synthetic derivatives of vinblastine used in combination chemotherapy of leukaemia, lymphoma, non-small cell lung cancer and non-metastatic breast cancer [21]. Although extensively studied, colchicine is not clinically used for cancer treatment because of its toxicity to normal cells. Numerous drugs, such as paclitaxel, docetaxel and polyisoprenyl benzophenones, are active tubulin polymerization enhancers. They stabilize microtubules and promote their polymerization. The paclitaxel-binding site on tubulin was characterized by Nogales et al. [22]. It is located on the inner surface of the microtubules in a deep hydrophobic pocket on -tubulin (Fig. 1). In tumor cells in vitro, paclitaxel potently suppresses the microtubule dynamic [15]. Paclitaxel binding to its site induces conformational changes in the tubulin structure, presumably increasing its affinity for neighboring subunits. Most microtubule-targeting drugs promote the apoptosis of cancer cells. However, the mechanism by which microtubule-interfering agents induce apoptosis is poorly understood. Generally, the action of MIAs causes the cells to arrest at the G2/M stage of the cell cycle. Li et al. [23] tried to explain the relationship between the mitotic block and the apoptotic signaling pathway caused by 2',3,4,4',5-pentamethoxy-trans-stilbene (PMS). They suggested that Bcl-2 and/or Bcl-xL hyperphosphorylation by CDC2 or JNK, BAD phosphorylation on serine 128 by CDK1, and Bim activation by JNK could be the mediators linking MIA-induced mitotic arrest and apoptosis. Moreover, it was postulated that MAPKs (ERK, p38, and JNK) did not play a critical role in the PMS-induced apoptotic response because the pharmacological inhibition of MAPKs did not affect the cytotoxicity and pro-apoptotic activity of PMS [23]. Activation of caspase-8 was observed during paclitaxel-induced apoptosis [24], but the relationship between the mitotic block and caspase-8 activation remains unknown. Regarding the mechanism of anti-carcinogenic action of MIAs, a strong correlation between the anti-tubulin activity of these agents and their vasculardisrupting action should be considered. In oncology, vascular-targeting interventions with the use of small molecule agents are believed to be efficient and worth further development. Compounds that interact with the tumor vasculature are known as vascular-targeting agents (VTAs) and are divided into two groups: a) angiogenesis-inhibiting agents (AIA) that prevent neovascularization in tumors and b) vascular-disrupting agents (VDAs) that influence already established tumor vessels [25]. Bevacizumab (Avastin), a humanized monoclonal antibody, is used in oncology as an anti-angiogenic VTA, but no VDAs have been approved by the US Food and Drug Administration (FDA) to date. The interest in the development of new VDAs is still growing and many VDAs are currently undergoing human clinical trials. There is a strong correlation between the vascular-disrupting activity of anticancer agents and their ability to inhibit tubulin assembly into microtubules in vitro [26]. By influencing tubulin polymerization, VDAs damage the proliferating endothelial cells of the cancer vasculature, thus decreasing the blood flow in the tumor tissue and leading to the death of cells in the central part of the tumor. Clinically relevant small-molecule VDAs interact with tubulin at the colchicine-binding site on -tubulin. The ability of the best known polymerizing agent, paclitaxel, to induce vascular damage has not yet been reported. VDAs are used in combination with standard radiotherapy and chemotherapy of solid tumors [27]. As angiogenesis also plays a crucial role in retinal and choroidal neovascularization leading to vision loss, preclinical studies have been performed on the potential curative activity of CA-4 in these diseases, and the results are promising [28][29][30].

STILBENE DERIVATIVES AS TUBULIN-INTERACTIVE AGENTS
Cancer chemotherapy agents should have good bioavailability, high cytotoxicity towards cancer cells, negligible cytotoxicity towards normal cells, and a wellrecognized potential for multi-targeted action. The search for new anticancer drugs is based on chemical modification of active compounds of natural origin or synthetic ones that show anti-carcinogenic activity in experimental model systems. Phenol compounds of plant origin, including cis-and trans-stilbene derivatives, are a very promising group of potential chemotherapeutic agents.
Combretastatin A4 and analogs CA-4 ( Fig. 3) was isolated from the root bark of the South African willow tree Combretum caffrum Kuntze [2]. Initially, the isolation was based on bioassay techniques that used the astrocyte reversal system. In 1987, further antineoplastic agents were isolated from this African tree (Fig. 3), namely combretastatin A-1 (CA-1), A-2 (CA-2), A-3 (CA-3), B-1 (CB-1) and B-2 (CB-2), and their potency as inhibitors of microtubule assembly was demonstrated [31,32]. In 1989, the cytotoxic activity of CA-4 and its anti-tubulin activity were described. It was found that CA-4 and CA-1 inhibit the growth of murine lymphocytic leukemia cells and human colon cancer cells [33]. Both CA-4 and CA-1 and their corresponding phosphate prodrug salts (CA-4P and CA-1P; Fig. 4) are of increasing therapeutic interest, but CA-4 seems to be more potent as a depolymerizing agent. Its anti-mitotic and cytotoxic properties have been reviewed by several authors [3,34,35]. Moreover, CA-4 shows a strong anti-angiogenic activity related to selective inhibition of the formation of new blood vessels in the cancer tissue [36,37]. CA-4 is one of the best-known inhibitors of tubulin polymerization in vitro. CA-4 binds to tubulin at the colchicine-binding site, leading to the formation of abnormal mitotic spindles as a result of the disturbance of the dynamic equilibrium of microtubule formation [35]. The IC 50 value (concentration that inhibits tubulin polymerization to 50%) of CA-4 was determined in several laboratories and ranged from 0.53 to 2.4 M [33,[38][39][40]. CA-4 exerts a potent cytotoxicity against many human cancer cell lines, including multi-drug resistant (MDR) cancer cells [35]. However, CA-4 does not display the expected activity in vivo due to its poor bioavailability caused by its low solubility and the instability of its cis conformation, which easily changes to form the trans-isomer [41]. Phosphates as prodrugs of combretastatins show better aqueous solubility and importantly, they are metabolized to their active forms by phosphatase, which exerts high activity in the tumor environment.  [34]. In clinical trials, a significant reduction in tumor blood flow was observed at doses equal to or below the maximum tolerated dose. In solid tumors, CA-4P exerted a therapeutic effect on the inner part of the tumor tissue, leaving a rim of viable tumor cells at the periphery. Therefore, CA-4P is proposed in combination with other therapies that are more effective at treatment of the outer tumor region [34,21]. The FDA has agreed to a Special Protocol Assessment (SPA) Phase III clinical trial (FACT 2) of Zybrestat for the treatment of highly aggressive and treatmentresistant anaplastic thyroid cancer. The trial will assess the effectiveness of CA-4P in combination with traditional chemotherapeutic agents or radiotherapy. The FACT 2 study is designed as a placebo-controlled, double-blind study of 300 subjects randomized 1:1 to receive carboplatin and paclitaxel plus CA-4 or carboplatin and paclitaxel plus a placebo (www.clinicaltrials.gov). CA-1 and its prodrug disodium CA-1P-2,3-diphosphate (CA-1P, OXi4503; Fig. 4) are cytotoxic against several human cancer lines [42] and the cell lining of tumor vasculature [43]. Oxi4503 is in several phase I/II clinical trials involving patients with solid tumors and the safety and dose-limiting toxicities of Oxi4503 are being estimated. The OXiGENE website, www.oxigene.com, is a good resource for current information on the progress of their clinical trials. To increase the bioavailability of CA-4, a bioreductively activated prodrug delivery system was proposed. A series of CA-4 nitrothiophene ether-linked conjugates was synthesized and a promising new lead was found. This compound was a gem-substituted derivative (dimethyl α-carbon substitution). It exerted aerobic metabolic stability and displayed the efficiency of hypoxiarelated release of CA-4 [44]. Investigations are ongoing into formulations that improve the bioavailability of MIAs, including nanoparticles, which are suggested to provide effective and selective delivery of cytotoxic agents to cancer cells [45]. Based on the results of preclinical studies, CA-4 derivatives were suggested to be effective for retinal neovascularization and choroidal neovascularization [28][29][30]46]. In 2007, positive phase II human trial results for ophthalmic formulations of CA-4 used against age-related macular degeneration were announced by OXiGENE. Information on currently ongoing clinical studies is presented at the official websites www.cancer.gov and www.clinicaltrials.gov. One second-generation CA-4-P analog that is suitable for therapeutic intervention and has entered clinical trials is a serinamide of 3'-amino-3,4,4', 5-tetramethoxy-cis-stilbene (ombrabulin; AVE8062; Fig. 4). Preclinical studies have proved the high anti-mitotic and vascular disrupting potency of this compound [47,48]. Concerning the synthesis and biological evaluation of structurally modified CA-4 analogs, it is worth referring to the latest studies of Pettit et al. They synthesized iodocombstatin and diiodocombstatin phosphate prodrugs in order to obtain an enhanced concentration of the drug in the thyroid carcinoma tissue. Both series of compounds displayed significant inhibition of the growth of human breast, central nervous system, lung and prostate cancer and leukemia cell lines. Two analogs (3-iodo-4,4',5-trimethoxy-3'-hydroxy-cis-stilbene and 3,5-diiodo-4,4'dimethoxy-3'-hydroxy-cis-stilbene) efficiently inhibited tubulin polymerization [49]. Summarizing, CA-4 exerts significant anti-mitotic and anti-angiogenic activities, and introduction of additional substituents to the CA-4 backbone is expected to improve its bioavailability and enhance its cytotoxic, anti-mitotic and proapoptotic activities. The achievements in this field have stimulated scientists to further their efforts to find new agents with clinical benefits.

Resveratrol analogs
Resveratrol (3,4',5-trihydroxy-trans-stilbene; 1; Fig. 5) is the best-known representative of the trans-stilbenes. It is a natural polyphenol with welldocumented anti-oxidative, anti-inflammatory and pro-apoptotic activity [50]. It occurs in the human diet as a constituent of peanuts, grapes and red wine. Its chemopreventive activity is supposed to be related to its ability to arrest cell cycle progression or to initiate tumor cell death by apoptosis [51]. Surprisingly, this compound does not interact with tubulin. The anti-proliferative activity of resveratrol may be explained by a mechanism that directly involves modulation of specific signaling pathways. However, both cis-and trans-stilbenes demonstrate anticancer properties in numerous experimental models in vitro and in vivo. It is assumed that the mechanism of their action is different. SAR analysis indicated the significance of the cis configuration of the olefinic bridge and 3,4,5-trimethoxyphenyl moiety in relation to the cytotoxic and anti-tubulin properties of stilbenes [3]. Substitution of the resveratrol molecule hydroxy groups with methoxy substituents improves the bioavailability of the compound by inhibiting conjugation reactions with sulphuric or glucuronic acids [51]. A methylated derivative of cis-resveratrol, 3,4',5-trimethoxy-cis-stilbene (Fig. 5, compound 2) displays 100-fold more potent anti-proliferative activity than the parent compound in completely arresting the growth of human colon cancer Caco-2 cells at 0.4 M. The cytotoxic effect of compound 2 is explained in terms of the depletion of the intracellular pool of polyamines and suppression of tubulin polymerization [52]. The compound only partially inhibited colchicine binding to tubulin, indicating that cis-resveratrol derivative binds to a specific site that is not identical with the colchicine-binding site, but colchicine binding may be influenced by allosteric interaction. In most studies, trans isomers of stilbene derivatives enhanced the process of tubulin polymerization, while cis isomers -analogs of CA-4 promoted the process of depolymerization, inhibiting the formation of microtubules. Mazue et al. [53] studied the anti-mitotic activity of a series of both cis-and trans-resveratrol analogs. Using a molecular docking method as a complement to experimental studies on the proliferation inhibition of the human colorectal tumor SW480 cell line, they showed that the cis configuration associated with the substitution of hydroxy groups with methoxy groups is crucial and increases inhibition efficacy. The docked structures of cis-polymethoxy isomers overlap with the docked structure of CA-4 at the tubulin colchicine-binding site, while transpolymethoxy isomers do not fit CA-4. However, isomers of both conformations induced polyploidy resulting from the blocking of cell divisions at the mitosis level [53]. 3,4,4',5-tetramethoxy-trans-stilbene (MR-4; DMU-212; 3; Fig. 5) is one of the most active resveratrol analogs that has been studied over the last ten years. It was reported to be 4 times more potent than resveratrol as an inhibitor of the growth of HCA-7 human colon cancer cells [54]. Moreover, it is also a potent inhibitor of adenoma development in the Apc Min+ mouse, a model of human intestinal carcinogenesis [55]. Other authors reported its higher cytotoxicity against two human breast cancer cell lines, MCF-7 and MDA-MB-435, in comparison with trans-resveratrol. The use of DMU-212 at a dose of 2.5 M resulted in a microtubule-stabilizing effect comparable to that of 10 M paclitaxel [56]. Another polymethoxy derivative of resveratrol, 2,3',4,4',5'-pentamethoxy-transstilbene (PMS; 4; Fig. 5) was found to be a potent inducer of apoptosis in colon cancer cells thanks to its targeting of microtubules [23]. However, for this trans isomer, enhancement of tubulin polymerization followed by G2/M mitotic arrest and caspase-dependent apoptosis was observed. PMS displayed a more potent anti-mitogenic effect on colon cancer cells (Caco-2, HT-29 and SW1116) when compared with 2',3,4',5-tetramethoxy-trans-stilbene (TMS; 5; Fig. 5). The potent cytotoxic effect of TMS was shown earlier using hormone-resistant breast cancer cells [57]. Recently, it was reported that PMS effectively suppressed colitis-associated colon carcinogenesis in mice (AOM/DSS animal model). The authors explained the chemopreventive effect of PMS in terms of suppression of cell proliferation, promotion of apoptosis, inactivation of -catenin and downregulation of iNOS [58].
Pettit et al. [88] synthesized a series of twenty-three -trans-nitrostyrenes. All but eight displayed inhibitory action towards tubulin polymerization with an IC 50 of less than 10 M. (4-benzyloxy-3-methoxyphenyl)-trans-nitrostyrene (Fig. 6, compound 10), with its bulky hydrophobic substituent, was the most active tubulin polymerization inhibitor in the series with an IC 50 of 2.5 µM. Its structure resembles the structure of combretastatins, in that there is a two-atom bridge between the two phenyl rings. Analogs of compound 10 with different substitution patterns in their phenyl rings seem to be worth further study. trans-3,4,5-trimethoxynitrostyrene was found to be a potent inhibitor with an IC 50 of 4.5 M, but it displayed relatively weak activity as an inhibitor of colchicine binding to tubulin. Moreover, it did not cause mitotic arrest in the mammalian cells that were treated with a cytotoxic concentration of the compound. It was suggested that a different cellular target is responsible for the cytotoxic activity of this nitrostyrene.

Tubulin-binding stilbenes as VDAs
CA-4 was the first small molecule VDA shown to have anti-vascular effects at low doses [89]. In tumors treated with CA-4-P, selective and extensive vasculature damage occurred, causing hemorrhagic necrosis and tumor growth delay [37]. Several CA-4 derivatives were described and patented as small molecule VDAs [42], including AVE 8062 and Oxi-4503 (Fig. 4), which demonstrate strong anti-tumor effects due to their vasculature-disrupting activity [90,91]. Ombrabulin enhances the efficacy of standard therapies (radiation plus cisplatin and radiation plus cetuximab) in head and neck squamous cell carcinoma (HNSCC) xenograft models. A specific action of ombrabulin towards intratumoral vessels rather than to the peritumoral vasculature was demonstrated by means of immunochemical staining of HNSCC tumors [92]. The design of CA-4 analogs as VDAs was the subject of a concise review by Rajak et al., who paid special attention to cis-restricted isomers [93]. Resveratrol's anti-angiogenic activity has been shown in in vitro assays and preclinical animal models [94,95]. However, this polyphenol is a weak tubulininterfering agent. Its anti-angiogenic activity is probably involved with the inhibition of the expression of HIF-1α and VEGF through multiple mechanisms, such as inhibition of Akt and MAP kinases, inhibition of protein translation regulators, and enhancement of proteosomal degradation of HIF-1α protein [96]. The anti-angiogenic and vascular-targeting activity of a series of transresveratrol derivatives was evaluated by Belleri et al. [97]. The prevention of new blood vessel formation and the disruption of the neovasculature by 3,4', 5-trimethoxy-trans-stilbene (Fig. 5, compound 6) was established with the use of a broad spectrum of in vitro and in vivo assays. Compound 6 exerted 30 to 100 times more potent anti-angiogenic activity in comparison to resveratrol in terms of inhibiting endothelial cell proliferation, sprouting, collagen gel invasion, and morphogenesis. In vivo, compound 6 caused the rapid stasis of blood flow and regression of intersegmental vessels in the trunk of zebrafish embryos [97].
More recently, these results were confirmed by Alex et al., and it was postulated that the anti-angiogenic and vascular-targeting activities of compound 6 result from the downregulation of VEGFR2 expression and cell-cycle arrest at the G2/M phase [98]. It remains unexplained why the trans-isomer of compound 6 affects tubulin polymerization in the same way as methoxystilbenes in the cis configuration, which share high structural similarity with CA-4. The tubulininterfering activity of trans-isomers that leads to cell cycle arrest [56,99] still needs to be elucidated. Cai et al. [62] synthesized six new stilbene derivatives and screened for cytotoxicity in different human tumor cell lines. They found that 3,4', 5-trimethoxy-cis-stilbene-3'-O-phosphate disodium (M410) is the most potent cytotoxic agent against multi-drug resistant tumor cells. M410 was a potent inhibitor of bovine brain tubulin polymerization in vitro. In experiments with human umbilical vein endothelial cells (HUVECs), rapid depolymerization of cellular microtubules, changes in cell shape and mitosis arrest were observed. In experiments on nude mice in vivo, M410 inhibited the growth of human colon carcinoma xenografts and reduced microvessel density in tumor tissues [62]. Most recently, it was shown that the cis isomer of CA-1P significantly impaired tumor blood flow leading to secondary tumor cell death and more than 95% tumor necrosis 24 h post drug exposure. It was suggested that CA-1P may be metabolized to orthoquinone intermediates leading to the formation of cytotoxic free radicals [99]. Treatment with the trans-isomer had no effect on the blood flow parameters. However, the combination of the trans-isomer with CA-4 increased the anti-tumor efficacy of the latter agent to near that of cis-CA-1P. These findings indicate that while the predominant in vivo effect of CA-1P is due to microtubule disruption and vascular shutdown, the formation of oxiquinones -toxic free radicals -could yield additional cytotoxic effects in vivo. Orthoquinones that bind to cellular nucleophiles form free radicals and have the potential to induce cancer cell death. Thus, the trans-CA-4P can contribute to the enhanced potency of the two isomers given to mice [100]. It has to be pointed out that the phosphorylated isomer did not show cytotoxicity in vitro, unlike the dephosphorylated orthoquinone species [101].

Molecular modeling and virtual screening
Molecular modeling makes an important contribution to the study of interactions between anti-mitotic agents and tubulin. The usefulness of this chemoinformatic approach for predicting the affinity of unknown compounds to molecular receptors has been confirmed in the virtual screening studies of tubulin inhibitors [102]. In order to create a tool for virtual screening, Nguyen developed a pharmacophore model [103] that was used and refined by other research groups [104]. Kim et al. screened a library of compounds with a 3,4,5trimethoxyphenyl motif selected from a chemical database [105]. Three compounds showed strong anti-proliferative activity against HL60 cell lines and were able to inhibit tubulin polymerization in vitro.
Recently, in the colchicine-binding domain, three zones of interaction were identified and specific amino acids responsible for the recognition and binding of known MIAs were determined [106]. Advanced in silico studies with the use of pharmacophore models were included in order to identify the most active cis-stilbene derivatives worthy of further synthesis. This strategy allowed a reduction in synthesis costs. However, in most of the studies, molecular docking is used to provide a rationale for experimental observations [61,107,108]. For example, Romagnoli et al. [107] synthesized and evaluated 2-amino-3-(3',4',5'-trimethoxybenzoyl)-6-substituted-4,5,6,7-tetrahydrothieno[2,3c]pyridine derivatives as anti-mitotic agents and inhibitors of tubulin polymerization. All of the compounds synthesized possessed a 3,4,5trimethoxyphenyl motif. The most potent inhibitors of tubulin polymerization were docked in the colchicine-binding site. Docking studies revealed that the trimethoxyphenyl moiety is situated in the same pocket on β-tubulin as the structurally analogous A ring of the co-crystallized DAMA-colchicine. The carbonyl group of the carbamate of 2-amino-3-(3',4',5'-trimethoxybenzoyl)-6ethoxycarbonyl-4,5,6,7-tetrahydrothieno[2,3-c]pyridine overlaps the carbonyl group of the C ring of DAMA-colchicine. 1,2,3-triazole analogs of CA-1 could be successfully docked into the colchicine-binding site of α,β-tubulin [82]. Kim et al. [109] synthesized a series of 43 stilbene derivatives and evaluated their cytotoxicity against human lung tumor cells (A5490). They used comparative molecular field analysis (CoMFA) to predict the biological activity of the designed molecules. With the use of the CoMFA pharmacophore model, they obtained a very good correlation between the calculated and observed IC 50 for the cytotoxic effect of synthesized compounds. Virtual screening is supposed to be a useful approach to discover new molecules with anti-proliferative activity through their ability to inhibit tubulin polymerization. Chiang et al. [110] identified novel tubulin inhibitors using a computational technique. Their complex approach included the generation and validation of a ligand-based pharmacophore model. This model was further employed for virtual screening of a chemical database containing approximately 130,000 compounds. Of the four compounds with anti-proliferative activity found in silico, one of them, compound 11 (Fig. 6), was identified as a potent anti-proliferative agent in KB cells with an IC 50 of 187 nM. Moreover, this indole inhibited tubulin polymerization with an IC 50 of 4.4 M and induced cell cycle arrest in the G2/M phase. Molecular docking revealed binding of compound 11 to the colchicine-binding site via two hydrogen bonds with Thr179 and Asn101 at distances of 2.89 and 3.05 Å, respectively. Interestingly, the chemical structure of this compound is distinct from colchicine and CA-4 analogs and it does not possess a trimethoxyphenyl moiety, which was reported as an important pharmacophore for the inhibition of tubulin assembly. However, its 2,5-dichlorobenzyl group occupies a similar region in tubulin as the trimethoxyphenyl of colchicine.
The challenge for further in silico studies will be the polymorphism of α-and β-tubulin subunits. For the β subunit, eight isoforms exerting 90% similarity have been identified and for the α subunit, at least six different isoforms are known [106]. They are characterized by different tissue distribution in normal cells, and in cancer cells, they may interact differently with chemotherapeutics. It is suggested that mutations leading to stabilization of microtubules may confer a resistance to drugs that act as destabilizers, such as vincristine and vinblastine, while the expression of isoforms causing destabilization of microtubules gives resistance to stabilizing agents, e.g. paclitaxel. Various isoform structures were obtained with the use of homology modeling. A set of novel colchicine derivatives with selective affinity to the colchicine domain of the αβIII isoform was synthesized. Their cytotoxic action correlated with the expression levels of specific tubulin isotypes [45,111]. The discovery of novel tubulin-binding agents that could selectively bind specific isoforms will be the aim of further research with the use of molecular modeling. Interesting research was reported on the poor affinity of paclitaxel to tubulin derived from the yew tree (Taxus baccata), which is a producer of this cytotoxin. The analysis indicated that the paclitaxel-binding site of yew tree tubulin contains several amino acid substitutions compared to human tubulin, all of which reduce the binding affinity of paclitaxel to yew tree tubulin [112].

SUMMARY
Intense research is being carried out to discover and elucidate the mechanism of action of anticancer therapeutic agents and to find the compounds that have superior therapeutic activity and bioavailability as future drugs. Tubulin is the focus of large-scale studies as a crucial target for anticancer drugs. The wellrecognized role of tubulin-interactive agents may help to develop new and efficient chemotherapeutics in the fight against cancer. The role of the combretastatin CA-4 and its numerous analogs seems to be significant. Taking into account the synergistic effect, the use of combretastatins in complex therapy is proposed in clinical trials (adjuvant therapy). Recently, it has been postulated that a multi-targeted management of tumors by combining antiangiogenic, vascular-disrupting and conventional chemotherapies and/or radiotherapy might be effective. Improvements in targeted drug delivery will contribute to the success of natural and synthetic products. Studies of the interaction between a drug and tubulin with the use of virtual docking and modeling create new perspectives for the design of synthetic analogs with improved properties and effectiveness in comparison to the parent compound.