ENMD-1198, a New Analogue of 2-Methoxyestradiol, Displays Both Antiangiogenic and Vascular-Disrupting Properties Free

Microtubule-targeting agents (MTA) represent a group of anticancer drugs that have been successfully used in the clinic for the treatment of a wide variety of cancers for more than 30 years (1). These chemotherapeutic drugs act by binding to α/β-tubulin heterodimers, resulting in either stabilization (taxanes) or depolymerization (Vinca alkaloids) of the microtubule network (2). This effect of MTAs on microtubules usually leads to mitotic arrest of treated cells at the metaphase/anaphase transition followed by apoptosis induction. Despite the extensive use of taxanes and Vinca alkaloids in the clinic, their efficacy is often limited by their toxic side effects, such as hematologic and neuronal toxicities. In addition, the anticancer activity of these drugs can be hampered by acquisition of the multidrug resistance phenotype, as they are substrates of efflux pumps, such as P-glycoprotein.

Despite their limitations, there is renewed interest in MTAs as a result of the discovery of their potent antivascular properties (3, 4). Almost 40 years after the pioneering work of Judah Folkman (5), tumor angiogenesis—the formation of new blood vessels from preexisting ones—has now become a major therapeutic target for the treatment of solid malignancies (6). Therapeutic strategies that target the tumor vasculature are extremely diverse and include anti–vascular endothelial growth factor (VEGF) antibodies, metalloproteinase inhibitors, receptor tyrosine kinase inhibitors, and conventional chemotherapeutic drugs. However, these strategies can be divided into two main categories: (a) antiangiogenic therapies that prevent the formation of a functional vascular network within the tumor and (2) vascular-disrupting therapies that destroy the established tumor vasculature. Noticeably, MTAs thus far represent the only class of anticancer drugs that can be successfully used in both types of antivascular strategies (4). For example, taxanes have been shown to inhibit tumor angiogenesis both in vitro and in vivo, whereas microtubule-depolymerizing agents, such as Vinca alkaloids and combretastatins, have been shown to disrupt the tumor vasculature (7).

The MTA 2-methoxyestradiol (2ME2), an endogenous metabolite of 17β-estradiol devoid of estrogenic activity, has been shown to have antiangiogenic and antitumor activity both in vitro and in vivo (8, 9). 2ME2 has been reported to bind to β-tubulin near the colchicine-binding site (10, 11). This binding results in kinetic stabilization of microtubule dynamics at low concentration (11) and inhibition of tubulin polymerization at higher concentrations, subsequently arresting the cell cycle at the G₂-M transition (10). 2ME2 has now moved to the clinic under the name Panzem (EntreMed), and encouraging results have been reported in clinical trials for the treatment of hormone-refractory prostate cancer (12), multiple myeloma (13), and, more recently, recurrent and platinum-resistant epithelial ovarian cancer (14). The advantage of 2ME2 over other MTAs is that it is not a substrate of multidrug resistance pumps (15) and it does not induce neurotoxicity or myelosuppression in cancer patients (12, 13, 16). However, most clinical trials also revealed that the bioavailability of 2ME2 was a limiting factor (12, 13, 17).

To improve the metabolic stability and antitumor efficacy of 2ME2, a series of analogues has been generated by substituent changes at the metabolically active positions 3 and 17 (18). From this series, three molecules were found to have improved metabolic stability and antiproliferative properties (19). One compound, ENMD-1198, was selected as the lead molecule of this series and is currently being tested in a phase I clinical trial in patients with refractory solid tumors. In the present study, we describe the potent antivascular properties of this new MTA. The effects of ENMD-1198 and 2ME2 on endothelial cell proliferation, motility, migration, VEGF receptor-2 (VEGFR-2) expression, and morphogenesis were analyzed using both a dermal [human microvascular endothelial cell line-1 (HMEC-1)] and a bone marrow–derived endothelial cell line (BMH29L). The capacity of these compounds to induce vascular disruption and the mechanisms involved, particularly their effects on the endothelial cytoskeleton and adhesion sites, were also investigated.

All cell lines were maintained in culture at 37°C and 5% CO₂. HMEC-1 cells were originally developed by Prof. Ades (Centers for Disease Control and Prevention, Atlanta, GA; ref. 20) and obtained from the Cell Culture Laboratory in the Hôpital de la Conception (Assistance Publique Hôpitaux de Marseille, Marseille, France). They were grown in MCDB-131 medium (Invitrogen) containing 10% heat-inactivated fetal calf serum (FCS), 2 mmol/L l-glutamine, 1% penicillin and streptomycin, 1 μg/mL hydrocortisone, and 10 ng/mL epithelial growth factor (BioScientific). BMH29L cells are bone marrow–derived endothelial cells that were immortalized by ectopic expression of human telomerase reverse transcriptase (21). They were kindly provided by Dr. Karen MacKenzie (Children's Cancer Institute Australia) and grown in Medium 199 (Invitrogen) containing 10% heat-inactivated FCS, 5% male human serum, AB only (Sigma-Aldrich), 1% penicillin and streptomycin, 1% heparin, 5 ng/mL recombinant human fibroblast growth factor β (Sigma-Aldrich), and 20 μg/mL endothelial cell growth factor (Roche). Both cell lines were cultured on 0.1% gelatin-coated flasks. MRC-5 fibroblast cell line was kindly provided by Dr. Belamy Cheung (Children's Cancer Institute Australia) and grown in MEM (Invitrogen) containing 10% FCS, 1% non-essential amino acids, 2% sodium bicarbonate, and 1% sodium pyruvate. All cell lines were regularly screened and are free from Mycoplasma contamination.

2ME2 and ENMD-1198 were gifts from EntreMed, Inc.; paclitaxel was purchased from Sigma-Aldrich; and vinblastine was from David Bull Laboratories. Stock solutions of 2ME2, ENMD-1198, and paclitaxel were stored in DMSO at −20°C. Vinblastine sulfate was kept in saline at 4°C.

Growth inhibition assays were done as previously described (22). Briefly, cells were seeded at 3,750 per well (HMEC-1 and MRC-5) or 1,500 per well (BMH29L) in 96-well gelatin-coated plates. After 24 hours, cells were treated with a range of concentrations of MTAs (vinblastine, paclitaxel, 2ME2, and ENMD-1198). After 72 hours of drug incubation, metabolic activity was detected by addition of Alamar blue and spectrophotometric analysis. Cell proliferation was determined and expressed as a percentage of untreated control cells. The determination of IC₅₀ values was done using GraphPad Prism 4 software (GraphPad Software, Inc.).

The Matrigel (BD Biosciences) assay was used to determine the antiangiogenic and vascular-disrupting properties of 2ME2 and ENMD-1198, as previously described (23). Briefly, 24-well plates were coated at 4°C with 270 μL of a Matrigel solution (1:1 dilution in MCDB-131 medium), which was then allowed to solidify for 1 hour at 37°C before cell seeding. For the antiangiogenesis analysis, cells were treated with different drug solutions 20 minutes after cell seeding, and photographs were taken after 6, 9, and 24 hours of drug incubation using the 5× objective of an Axiovert 200M fluorescent microscope coupled to an AxioCamMR3 camera driven by the AxioVision 4.7 software (Carl Zeiss). For the vascular disruption analysis, cells were allowed to undergo morphogenesis and form capillary-like structures for 6 hours before drug treatment was initiated. Photographs were then taken using the same microscope device after 2 and 18 hours of drug incubation. The antiangiogenic and vascular-disrupting activities of the compounds were then quantitatively evaluated by measuring the total surface area of capillary tubes formed in at least 10 view fields per well using AxioVision 4.7 software. Time-lapse videomicroscopy experiments were also done to illustrate the antivascular properties of 2ME2 and ENMD-1198. For antiangiogenesis assays, photographs were taken every 5 minutes from the beginning of drug treatment and for 9 hours. For vascular disruption assay, photographs were taken every 2 minutes from the beginning of drug treatment and for 2 hours. In both cases, cells were constantly maintained at 37°C and 5% CO₂.

The effect of 2ME2 and ENMD-1198 on VEGFR-2 expression was assessed by immunoblot, as previously described (24). Briefly, cells were seeded on a 100-mm Petri dish and treated 24 hours later with 2ME2 or ENMD-1198 in the presence or absence of FCS. After 24 hours of drug incubation, adherent cells were harvested in PBS using a cell scraper, pooled with floating cells, centrifuged to a pellet, and lysed in radioimmunoprecipitation assay buffer. Equal amounts of protein (35 μg) were resolved on 10% SDS-PAGE before electrotransfer onto nitrocellulose membrane. Immunoblotting was done using anti–VEGFR-2 (clone 55B11; Cell Signaling Technology) and anti–glyceraldehyde-3-phosphate dehydrogenase (GAPDH; Abcam) antibodies. The membranes were then incubated with horseradish peroxidase–conjugated IgG secondary antibodies, and protein was detected with ECL Plus (GE Healthcare Life Sciences). The blots were scanned and densitometric analysis was done as previously described (25).

Random cell motility was assessed by time-lapse microscopy as previously described (26). Briefly, cells were seeded on a 24-well gelatin-coated plate and allowed to adhere for 1 hour. The cells were then incubated with 2ME2 and ENMD-1198, and photographs were taken every 5 minutes for 6 hours in at least two view fields per well using the 5× objective of the same microscope device used for Matrigel assays. Analysis was done using the tracking module of the AxioVision 4.7 software. At least 25 cells per view field were tracked for 6 hours; cells undergoing division or apoptosis were excluded from analyses. The persistent random walk model was used to characterize cell motility (27). For each individual cell, the mean square displacement <D²> was calculated from the following formula:

where d_i is the displacement of a cell from its initial position at time level t_i. The random motility coefficient (μ), which reflects the rate at which a cell population is able to migrate into and colonize a new area, and the persistence time (P), which represents the average time interval between significant direction changes, were deduced from the <D²> value using the following formula:

where t is time, S is the average cell speed, and n = 2 in our two-dimensional walk model.

The wound-healing assay was done as previously described (28) with slight modifications. Specifically, cells were seeded on 60-mm gelatin-coated Petri dishes with 2-mm grids (Corning). When the cells reached 70% to 80% confluence, the culture medium was replaced with either 2ME2 or ENMD-1198 solution at varying concentrations. After 2 hours of drug incubation, four scratches (wounds) were made on the monolayer with a pipette tip and cells were washed with fresh medium to remove cell debris. Photographs of six view fields on each plate were then taken at marked positions on the wound at 0-, 2-, 4-, 6-, 8-, and 24-hour time points, with the 5× objective of an Axiovert S100 microscope (Carl Zeiss) coupled to a SPOT RT Slider Camera with SPOT software Windows version 4.0.2 (Diagnostic Instruments). The wound area at each position was then measured using ImageJ 1.37 software, and the percentage of wound closure at each time point was calculated as the percentage of the wound area at a given time point relative to the wound area at time 0.

The chemotaxis assay was done according to Brown et al. (29) with slight modifications. Specifically, the underside of 8-μm transparent polyethylene terephthalate membrane inserts for 24-well plates (BD Falcon) was precoated with 0.1% gelatin for 1 hour. The cells were prelabeled in situ with 10 μmol/L CellTracker Green CMFDA (Invitrogen) in serum-free medium (MCDB-131, Medium 199, and MEM for HMEC-1, BMH29L, and MRC-5 cells, respectively) for 30 minutes, and 100,000 cells were then seeded onto the insert in drug-containing assay medium (0.5% bovine serum albumin in serum-free medium). Drug-containing assay medium with 5% FCS was then added to the bottom of the insert and used as chemoattractant. A negative control was included in each experiment by adding serum-free assay medium to the bottom of the insert. The plates were incubated for 6 hours at 37°C and 5% CO₂. Excess cells on the upper side of the insert were then gently swabbed off with a cotton tip, and migrated cells at the underside of the insert were measured with a Victor 3 plate reader (Perkin-Elmer) at 492/517 (absorption/emission). All readings were then normalized to the positive control (drug-free 5% FCS assay medium).

The effect of 2ME2 and ENMD-1198 on microtubules, actin filaments, and adhesion sites was evaluated by immunofluorescence. Briefly, endothelial cells were seeded on gelatin-coated 8-well Permanox Lab-Tek chamber slides (Life Technologies). The medium was then replaced after 24 hours with drug solution at varying concentrations and incubated for 2 hours. Tubulin was stained as described by Don et al. (30) with slight modifications. Specifically, cells were fixed and permeabilized with 100% methanol for 15 minutes at −20°C and blocked for 10 minutes with 10% FCS/PBS. α-Tubulin was then probed with an anti-mouse α-tubulin primary antibody (1:500 in 5% FCS/PBS; Sigma-Aldrich) for 1 hour followed by Alexa Fluor 488 anti-mouse secondary antibody (1:250 in 5% FCS/PBS; Invitrogen) for 1 hour. Actin cytoskeleton staining was done as previously described (31) with a few modifications. Cells were fixed with 3.7% formaldehyde/PBS for 10 minutes and permeabilized with 0.1% Triton X-100/PBS for 3 minutes. Filamentous actin was then stained with Alexa Fluor 568 phalloidin (5 μg/mL in 5% FCS/PBS) for 10 minutes, and focal adhesions were stained with either vinculin (Sigma-Aldrich) or paxillin (BD Biosciences) primary antibody (1:1500 in 5% FCS/PBS) followed by Alexa Fluor 488 anti-mouse secondary antibody (1:1,000 in 5% FCS/PBS). All slides were mounted on coverslips with ProLong Gold antifade reagent containing 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen) and imaged with the same microscope device used for Matrigel assays.

All experiments were done at least in triplicate. Statistical significance was determined using two-sided unpaired Student's t test in the GraphPad Prism 4 software.

The effect of 2ME2 and the analogue ENMD-1198 on the proliferation of two independent endothelial cell lines, HMEC-1 and BMH29L, was investigated by Alamar blue assay using drug solutions ranging from 0.1 and 20 μmol/L. In both cell lines, ENMD-1198 was more potent than 2ME2 at inhibiting endothelial cell proliferation (Fig. 1). After 72 hours of incubation, the IC₅₀ values were 2.3 and 0.4 μmol/L in HMEC-1 cells for 2ME2 and ENMD-1198 cells, respectively (Fig. 1A; P < 0.01), whereas they were 8.2 and 3.8 μmol/L in BMH29L cells (Fig. 1B; P < 0.05). The higher sensitivity of HMEC-1 cells to 2ME2 and ENMD-1198 compared with BMH29L cells was also reflected with the other MTAs tested (paclitaxel and vinblastine; Supplementary Table S1). These results show that ENMD-1198 exerts a stronger antiproliferative activity in endothelial cells than its parental compound 2ME2.

To investigate the capacity of ENMD-1198 to inhibit angiogenesis in vitro, the Matrigel assay was used. Time-lapse videomicroscopy revealed that ENMD-1198 prevents vascular endothelial cells from forming capillary-like structures (Supplementary Videos S1 and S2). Incubation of HMEC-1 cells with concentrations of drug at the IC₅₀ for cell proliferation—0.5 and 2 μmol/L for ENMD-1198 and 2ME2, respectively—inhibited the formation of capillary tubes by ∼70% (Fig. 2A). Dose-response experiments showed that ENMD-1198 was more potent than 2ME2 at inhibiting endothelial cell morphogenesis in vitro (Fig. 2B). Indeed, ENMD-1198 and 2ME2 significantly inhibited capillary tube formation from 0.1 and 1 μmol/L, respectively. In addition, endothelial cell morphogenesis was completely suppressed by 1 and 5 μmol/L of ENMD-1198 and 2ME2, respectively. Similar results were obtained when analyzing the number of closed vascular structures and the total length of tubes per view field and when experiments were done on BMH29L cells (data not shown).

The effect of 2ME2 and ENMD-1198 on VEGFR-2 expression was evaluated by Western blotting. Interestingly, when control HMEC-1 cells were serum deprived for 24 hours, an increase in the expression level of VEGFR-2 was observed (47 ± 8%; P < 0.05; Fig. 2C). Incubation for 24 hours with 2ME2 at IC₅₀ for cell proliferation did not significantly affect the basal expression level of VEGFR-2 but partially inhibited its upregulation induced by serum deprivation (P < 0.05). However, incubation with ENMD-1198 at the IC₅₀ for cell proliferation caused a significant decrease in VEGFR-2 expression both in the presence (−26 ± 3%; P < 0.01) and in the absence of serum (−46 ± 7%; P < 0.01). Altogether, these results showed that ENMD-1198 displayed more potent antiangiogenic properties than 2ME2.

To better understand the antivascular properties of ENMD-1198 and its parental compound 2ME2, we assessed their effect on endothelial cell random motility by time-lapse videomicroscopy experiments using the persistent random walk model (27). HMEC-1 cells were seeded on gelatin-coated plates, and photographs were taken every 5 minutes for 6 hours to track cell movements. Figure 3A shows the representative trajectories of five cells, either untreated (left) or treated with drug at a concentration equivalent to the IC₅₀ for cell proliferation (ENMD-1198, middle; 2ME2, right). Both drugs significantly inhibited endothelial cell motility. Cells treated with ENMD-1198, and to a lesser extent with 2ME2, remained focused in a smaller area than the control untreated cells (Fig. 3A). Analysis using the random walk model (see Materials and Methods) showed that both ENMD-1198 and 2ME2 had an effect on the average cell velocity (S), the persistence time (P; average time between significant direction changes), and the random motility coefficient (μ), which reflects cell ability to migrate and colonize a new area (Fig. 3B and C; additional data not shown). The average velocity of HMEC-1 cells (0.74 ± 0.03 μm/min) was significantly decreased by 27.5% and 22% when cells were treated with the IC₅₀ of ENMD-1198 and 2ME2, respectively (Fig. 3B; P < 0.01 for both drugs). The average persistence time of HMEC-1 cells (257 ± 7 s) was significantly increased by 20% and 18% when the cells were treated with the IC₅₀ of ENMD-1198 and 2ME2, respectively (Fig. 3C; P < 0.01 for both drugs). Incubation of HMEC-1 cells with the same drug concentrations also resulted in a 38% and 31% decrease in the random motility coefficient (μ) with ENMD-1198 and 2ME2, respectively (data not shown; P < 0.01 for both drugs). Dose-response experiments showed that the effect of these drugs on endothelial cell motility was dose dependent and that ENMD-1198 was more potent at inhibiting endothelial cell motility than the parent compound (Fig. 3B and C). ENMD-1198 was able to significantly inhibit endothelial cell motility at 0.5 μmol/L, whereas 2 μmol/L concentration of 2ME2 was required for significant inhibition (Fig. 3B and C). In addition, ENMD-1198 induced a more dramatic inhibition of endothelial cell motility than 2ME2 at all concentrations tested. Experiments done in BMH29L cells confirmed the inhibitory effects of both drugs on cell motility and the more potent effects of ENMD-1198 compared with 2ME2 (data not shown).

To further investigate the effect of ENMD-1198 and 2ME2 on the capacity of endothelial cells to migrate, wound-healing experiments were done. Figure 4A shows representative photographs 8 hours after the wound was inflicted to the endothelial cell monolayer. At this time point, recovery from the wound was 62.4 ± 4.8% in control untreated HMEC-1 cells, whereas it was only 33.7 ± 1.6% and 48.8 ± 4.8% for cells treated with 0.5 μmol/L ENMD-1198 and 2 μmol/L 2ME2, respectively (P < 0.05 for both drugs). At all time points (2, 4, 6, 8, and 24 h), both drugs significantly inhibited wound recovery (P < 0.05; data not shown). Linear regression analysis showed that control cells reached 50% wound recovery after 6.8 ± 0.8 hours, whereas cells treated with ENMD-1198 and 2ME2 reached 50% wound recovery after 13.5 ± 2.2 hours and 8.5 ± 0.9 hours, respectively (P < 0.05).

In addition to wound-healing experiments, chemotaxis assays were also done using the modified Boyden chamber assay (see Materials and Methods). As shown in Fig. 4B, addition of 5% FCS in the bottom well of the inserts resulted in a 3-fold increase in HMEC-1 migration during a 6-hour incubation period compared with the negative control (absence of FCS). This increase in HMEC-1 cell migration was significantly inhibited by concentrations of ENMD-1198 ≥1 μmol/L or 5 μmol/L for 2ME2. Dose-response experiments showed that inhibition of endothelial cell chemotaxis was dose dependent and that ENMD-1198 was more potent at inhibiting endothelial cell migration than 2ME2 (Fig. 4B). Similarly in BMH29L cells, ENMD-1198 was able to significantly inhibit chemotaxis from 2 μmol/L, whereas 10 μmol/L concentration of 2ME2 was required (data not shown).

To investigate the potential changes in the endothelial cytoskeleton associated with the antivascular activity of ENMD-1198, we did immunofluorescence experiments. Staining with α-tubulin antibody revealed that ENMD-1198 induced rapid and extensive depolymerization of the microtubule network in vascular endothelial cells at concentrations ≥0.5 μmol/L (Fig. 5A). Interestingly, 2ME2 also induced significant depolymerization of microtubules from 2 μmol/L, albeit to a lesser extent than ENMD-1198 (Fig. 5A). We then analyzed the effect of microtubule depolymerization on the actin cytoskeleton and adhesion sites by doing dual staining with phalloidin and paxillin antibody. Consistent with complete microtubule depolymerization induced by ENMD-1198, an accumulation of actin stress fibers was observed (Fig. 5B, top, arrowheads). In accordance with the results of microtubule staining, treatment with 2ME2 also resulted in actin stress fiber accumulation but to a lesser extent than observed with ENMD-1198 (Fig. 5B, top). Finally, treatment with either ENMD-1198 or 2ME2 also altered the organization of adhesion sites (Fig. 5B, middle). Paxillin staining showed that control untreated cells mainly display small adhesion complexes, most of them located in the lamellipodia (Fig. 5B, middle, arrows). Incubation with either ENMD-1198 or 2ME2 inhibited the formation of membrane protrusions such as lamellipodia and led to an accumulation of large and disorganized focal adhesions throughout the cell periphery (Fig. 5B, middle).

The capacity of ENMD-1198 to rapidly induce microtubule depolymerization in endothelial cells suggested that this compound could exhibit potent vascular-disrupting properties in addition to its antiangiogenic properties. To investigate the potential of ENMD-1198 to disrupt preformed vascular structures, endothelial cells were first allowed to form capillary-like tubes on Matrigel before treatment initiation. After 6 hours of incubation in the absence of drug, drug solution was added and time-lapse videomicroscopy was done. As shown in Supplementary Data, ENMD-1198 induced complete collapse of the vascular network within 2 hours of drug treatment (Supplementary Videos S3 and S4). ENMD-1198 displayed more potent vascular-disrupting properties than 2ME2 (Fig. 6). Indeed, dose-response experiments showed that ENMD-1198 and 2ME2 significantly disrupted vascular structures in 2 hours from 0.25 and 2 μmol/L, respectively (Fig. 6B). Moreover, incubation of HMEC-1 cells for 2 hours with drug at a concentration equivalent to the IC₅₀ for cell proliferation resulted in a more extensive disruption of capillary-like structures with ENMD-1198 compared with 2ME2 (48% and 29%, respectively; P < 0.01). The same difference of vascular-disrupting activity was observed between the two compounds when photographs were taken after 18 hours of drug incubation, when analyzing the number of closed vascular structures and the total length of capillary tubes per view field, and when experiments were done on BMH29L cells (data not shown).

Despite the clinical success of taxanes and Vinca alkaloids, numerous new MTAs are currently being developed to improve treatment efficacy, overcome drug resistance, and decrease toxic side effects. An endogenous metabolite of estradiol, 2ME2 (Panzem), which displays microtubule-destabilizing properties, has shown encouraging results for the treatment of multiple myeloma and prostate and ovarian cancer (12, 14). However, its bioavailability seems to be a limiting factor (12, 13, 17). To improve the metabolic stability and antitumor efficacy of 2ME2, a series of analogues has been generated (18, 19). The most promising of these analogues, ENMD-1198, has been selected as a lead compound and is currently being tested in phase I clinical trials in patients with refractory solid tumors. In preclinical studies, ENMD-1198 retained potent antitumor properties while displaying improved metabolic stability (19). In the present study, we have shown that ENMD-1198 is a potent antivascular agent.

ENMD-1198 was 2.5- to 6-fold more potent than 2ME2 at inhibiting human endothelial cell proliferation depending on the concentrations compared and the cell type. These results are consistent with those obtained with the fibroblast cell line MRC-5 (Supplementary Table S1) and those previously reported for another endothelial cell type (human umbilical vascular endothelial cell) and a variety of human cancer cell lines (19). Furthermore, our results show that, in addition to its improved antiproliferative activity, ENMD-1198 also more potently inhibits endothelial cell motility, chemotaxis, and morphogenesis into capillary-like structures. Endothelial cell motility is of critical importance in tumor angiogenesis especially in capillary growth and vascular network structure (32). Treatment with ENMD-1198 decreased the mean velocity of endothelial cells as well as their random motility coefficient (μ), which reflects the capacity of cells to explore and colonize a new area, and increased the time between significant direction changes (P). It also decreased the capacity of endothelial cells to respond to chemoattractive signals. Results of chemotaxis assay done with the fibroblast cell line MRC-5 suggested that the inhibition of migration by ENMD-1198 may not be specific to vascular endothelial cells (data not shown). Finally, ENMD-1198 also blocked the morphologic differentiation of endothelial cells into vascular structures on Matrigel. Thus, our study shows that ENMD-1198 can disrupt most endothelial functions involved in tumor angiogenesis.

Interestingly, Moser et al. (33) recently showed that ENMD-1198 could disrupt hypoxia signaling in hepatocellular carcinoma cells both in vitro and in vivo. In particular, treatment with ENMD-1198 prevented the accumulation of hypoxia-inducible factor-1α induced by hypoxic conditions in these cells, thus blocking the hypoxia-induced expression and secretion of VEGF (33). Here, we showed that ENMD-1198 could also directly decrease the expression level of VEGFR-2 in endothelial cells. This result, together with the study done on hepatocellular carcinoma cells (33), shows that ENMD-1198 can affect tumor angiogenesis at two different levels: at the tumor level by blocking hypoxia-inducible factor-1α signaling in cancer cells and at the endothelium level by blocking endothelial cell functions involved in angiogenesis and by decreasing VEGFR-2 expression. Moreover, treatment of mice bearing subcutaneous hepatocellular carcinoma with ENMD-1198 reduced tumor growth as well as the number of CD31⁺ blood vessels inside the tumors (33). Altogether, these data suggest that the in vivo antitumor activity of ENMD-1198 is at least in part mediated by its antiangiogenic activity.

The present study is the first report to show that 2ME2 and the analogue ENMD-1198 exhibit potent vascular-disrupting activity. Time-lapse videomicroscopy revealed that ENMD-1198 and, to a lesser extent, 2ME2 were able to induce rapid (30 min) disruption of preformed vascular structures. Interestingly, when comparing equitoxic concentrations, paclitaxel displayed similar antiangiogenic activity to ENMD-1198 and 2ME2 (Supplementary Fig. S5A). In contrast to ENMD-1198, paclitaxel had no detectable vascular-disrupting activity (Supplementary Fig. S5B). This result confirms that the rapid vascular disruption induced by ENMD-1198 is not due to its cytotoxic activity and that although almost all MTAs exhibit potent antiangiogenic properties, only MTAs with microtubule-depolymerizing properties can induce vascular disruption. Similarly to another class of microtubule-depolymerizing agents, combretastatins (31), the vascular-disrupting activity of ENMD-1198 and 2ME2 was associated with extensive microtubule depolymerization and accumulation of actin stress fibers. In addition, at concentrations that induce microtubule depolymerization, we observed a decrease in membrane protrusions, such as lamellipodia, and an accumulation of large focal adhesions. The newest Vinca alkaloid, vinflunine, was also shown recently to inhibit endothelial cell motility, in part by disrupting the cross-talk between microtubules and adhesion sites (34). Thus, rather than a direct effect, ENMD-1198 may induce the accumulation of large focal adhesions by inhibiting the microtubule-mediated regulation of adhesion site dynamics.

Altogether, our results show that ENMD-1198 is a potent antivascular agent compared with its parental compound 2ME2. Clinical evaluation of 2ME2 has reported promising results for the treatment of various human cancers. The dual-acting nature of ENMD-1198, as both an antivascular and an anticancer agent, strongly suggests that ENMD-1198 may prove to be a promising agent in the clinic.

No potential conflicts of interest were disclosed.

We thank Dr. Karen MacKenzie for reviewing our manuscript and Drs. Tony Treston and Mark Bray (EntreMed) for supplying the 2ME2 and ENMD-1198 used in this study.

Grant Support: Children's Cancer Institute Australia for Medical Research (which is affiliated with the University of New South Wales and Sydney Children's Hospital), Cancer Institute New South Wales (E. Pasquier), and Cancer Council New South Wales (M. Kavallaris). E. Pasquier is supported by a Cancer Institute New South Wales “Early Career Development” Fellowship and M. Kavallaris is supported by a National Health and Medical Research Council Senior Research Fellowship.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.