Structure Activity Analysis of 2-Methoxyestradiol Analogues Reveals Targeting of Microtubules as the Major Mechanism of Antiproliferative and Proapoptotic Activity Free

2-Methoxyestradiol (2ME2) is a naturally occurring metabolite of 17β-estradiol with antiproliferative, antiangiogenic, and proapoptotic activity (1, 2), and is considered to have potential clinical benefit in the treatment of cancer. 2ME2 does not interact with the estrogen receptors. Its major target is believed to be the cellular microtubule skeleton. 2ME2 has been shown to inhibit the polymerization of tubulin in vitro, thus disrupting normal microtubule function (3). However, at therapeutic doses in vivo, the drug is likely to act by impairing microtubule dynamics (4, 5). Given that rapid microtubule dynamics are essential for the correct assembly and function of the mitotic spindle, 2ME2 activates the spindle assembly checkpoint and causes metaphase arrest, and as a consequence, inhibition of cell proliferation and induction of cellular apoptosis (6, 7).

In addition to the inhibition of microtubule dynamics and induction of cell cycle arrest, the proapoptotic action of 2ME2 has also been linked to the stimulation of cellular reactive oxygen species (ROS) production, resulting in the release of cytochrome c from the mitochondria and activation of caspases (8–13). Specifically, 2ME2 has been reported to inhibit superoxide dismutase enzymes (8). However, a subsequent study documented that 2ME2-induced superoxide production is not due to superoxide dismutase inhibition and suggested the existence of an alternative mechanism (14). We and others have shown that 2ME2 and related compounds are inhibitors of complex I of the mitochondrial electron transport chain (15, 16). Given that in the majority of cell types, the mitochondrial electron transport chain is the major source of cellular ROS, this drug target may be responsible for 2ME2-induced superoxide production. Other complex I inhibitors, such as rotenone or 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine, have been shown to cause the production of ROS and induce cellular apoptosis, leading to neuronal cell death (17). Thus, 2ME2-dependent disruption of mitochondrial function may be an alternative mechanism for its proapoptotic effects.

2ME2 has also been reported to inhibit hypoxia-inducible factors (HIF; ref. 18). HIF is a transcription factor responsible for mediating the adaptation of cells and tissues to low oxygen concentrations by upregulating genes involved in angiogenesis and glycolysis (19). HIF is a heterodimer comprised of an oxygen-regulated α subunit (HIF-1α or HIF-2α) and a constitutively expressed β subunit (HIF-1β). The oxygen-dependent regulation of HIF-1α and HIF-2α occurs at the level of their protein stability. HIF-1α and HIF-2α are normally constitutively hydroxylated at two proline residues by oxygen- and oxoglutarate-dependent prolyl 4-hydroxylases (20, 21). Prolyl hydroxylation targets these proteins for ubiquitination by the pVHL-associated E3 ubiquitin ligase, resulting in 26S proteasome-mediated degradation. At low oxygen concentrations, prolyl hydroxylase activity is inhibited, resulting in HIF-1α and HIF-2α protein accumulation. 2ME2 was shown to inhibit hypoxia-induced HIF-1α protein accumulation (18). This effect was reported to be independent of oxygen-regulated HIF-1α protein stability, but due to inhibition of new HIF-1α protein synthesis. Inhibition of HIF-1α translation by 2ME2 was found to be caused by the disruption of microtubules, although the exact mechanism is currently unclear.

It has been suggested that the antiangiogenic effect of 2ME2 is a consequence of HIF inhibition (18). Various studies have also shown the importance of HIF-1α in cell proliferation in vitro and in vivo. Thus, knockout of HIF-1α resulted in the inhibition of the proliferation of colon cancer cells (22, 23), mouse embryonic stem cells (24), and immortalized mouse embryonic fibroblasts (25). HIF-1α deficiency has also been reported to promote apoptosis in mouse embryonic fibroblasts and pancreatic cancer cells (26, 27). However, in other models, such as in xenografted embryonic stem cells, HIF-1α increased apoptosis (24). Given the role of HIF-1α in cell proliferation and apoptosis, it is possible that the antiproliferative and proapoptotic effects of 2ME2 are mediated via HIF-1α inhibition. However, there are contradictory results in the literature as to whether 2ME2-mediated HIF-1α inhibition occurs at therapeutic drug concentrations.

In an effort to gain a better understanding of the contribution of the different 2ME2 targets to antiproliferative and proapoptotic drug effects, we used a number of different structural analogues and correlated their cellular apoptosis-inducing effects with their effects on different drug targets. Our results indicate that targeting of microtubules is the major mechanism of action of 2ME2, whereas induction of cellular ROS or disruption of mitochondrial function are unlikely to be important contributors to drug action in epithelial cancer cells. Inhibition of HIF-1α requires supratherapeutic concentrations and does not correlate well with effects on microtubules. Our studies also reveal interesting differential structural requirements for the effects on different drug targets.

MTT assays were done as previously described (28). GI₅₀ values were calculated with GraphPad Prism using nonlinear regression (sigmoidal dose-response variable slope curve fit).

MCF7 and HCT116 cells were grown on glass coverslips and treated with 25 μmol/L of 2ME2 analogues for 12 h. Cells were fixed using 4% formaldehyde and permeabilized with 0.2% Triton X-100 for 10 min. After blocking with 5% fetal bovine serum, the coverslips were incubated with mouse anti–α-tubulin followed by goat anti-mouse secondary antibody tagged with FITC. Coverslips were mounted onto glass slides using Vectashield (Vector Laboratories) containing 4′,6-diamidino-2-phenylindole to label nuclei.

Cell lysis and Western blotting was carried out as previously described (29). The following antibodies were used: mouse anti–α-tubulin (BD Transduction Laboratories), mouse anti–HIF-1α (BD Transduction Laboratories), mouse anti-FLAG (Sigma), rabbit anti–poly(ADP)-ribose polymerase (PARP; Cell Signaling Technology), mouse anti-p53 (Sigma), and rabbit anti-catalase (Abcam). The Western blots shown are representative of at least two independent experiments.

Cells in 12-well plates were transfected for 24 h with 0.2 μg of pGL3-HRE plasmid containing the hypoxia response element (HRE) from the phosphoglycerate kinase promoter (a kind gift from Dr. Kaye Williams). Cells were treated with DMSO or 25 μmol/L of 2ME2 analogues and incubated under normoxic or hypoxic conditions. Luciferase reporter activity was measured using the Steady-Glo luciferase assay reagent (Promega).

The T-Rex system (Invitrogen) was used to generate a cell line with tetracycline-inducible expression of P402A/P564A mutant full-length HIF-1α-FLAG (COOH-terminal 2× FLAG tag). Single clones were isolated using zeocin selection. Tetracycline-dependent protein expression was confirmed by immunoblotting.

Superoxide production was estimated using dihydroethidium. Cells were plated in 12-well plates and grown to 70% confluence and cotreated with dihydroethidium and 2ME2 analogues for 1 h. Subsequently, the cells were rinsed with PBS and kept in PBS for analysis of fluorescence using a multiwell plate reader (Molecular Devices), excitation and emission wavelengths were 520 and 610 nm, respectively.

Murine catalase was PCR-amplified from I.M.A.G.E. clone (ID 4241333) and inserted in the retroviral expression vector Puro-MaRX. It has been previously shown that a robust overexpression of catalase results in the accumulation of protein in the cytoplasm. The NH₂-terminal mitochondrial targeting sequence of SOD2 (comprised of the sequence corresponding to residues 1–28) was used to target catalase to the mitochondria. This construct also included a COOH-terminal V5 epitope tag. Retroviral particles were generated using catalase, mitochondrial catalase-V5, or enhanced green fluorescent protein (control) containing Puro-MaRX retroviral expression vector in 293-gag-pol cells and pseudotyped with vesicular stomatitis virus-g protein. Protein expression was assessed by fluorescence (EGFP) or immunoblotting with catalase antibody.

Mitochondria were isolated from mouse liver by differential centrifugation in mitochondria isolation buffer containing 280 mmol/L of sucrose, 10 mmol/L of Tris (pH 7.4), and 1 mmol/L of EDTA. Oxygen consumption was measured with a Clark-type oxygen electrode (Oxygraph, Hansatech) in mitochondria isolation buffer using 0.5 mg/mL of mitochondria with substrates and other additions as indicated in the figure legends.

2ME2 and the analogues used were obtained from EntreMed, Inc. The synthesis of the analogues is described in a series of articles which have been submitted for publication. The preclinical activity of ENMD-1198, which is currently being tested in a phase 1 oncology trial in patients with solid tumors, has been described in refs. (30, 31).

To determine the effects of the various 2ME2 analogues on cell proliferation, MTT assays were done in the colon cancer cell line HCT116 and the breast cancer cell line MCF7. In HCT116 cells, 2ME2 had the strongest inhibitory effect with a GI₅₀ value of 0.8 μmol/L, followed by ENMD-1269, ENMD-1198, and ENMD-1189 (see Fig. 1). In MCF7, the greatest inhibition was observed with ENMD-1198 (GI₅₀ of 0.4 μmol/L), followed by 2ME2, ENMD-1269, and ENMD-1189. ENMD-1151, which has a 2-amino group, but is otherwise identical to 2ME2, had only small antiproliferative or cytotoxic effects in the MTT assays in both cell lines, confirming the importance of the 2-methoxy group. Similarly, ENMD-1147, which is identical to 2ME2 except for the presence of a 17-hydroxy group in the α-configuration, had low activity compared with 2ME2. This indicates that the modification in the 17-position is of great importance for drug activity and that both 17β-hydroxy and 16,17-olefin groups confer activity. Finally, ENMD-1506, which has a 3-carboxy group, exhibited no inhibitory effect. This may be due to the inactivity of this analogue. However, because the drug also had no activity in any of the other cellular assays (see below), it seems likely that the 3-carboxy group prevents efficient uptake by cells. Our previous studies have shown that the modification at the 3-position is of great importance for drug activity, as an analogue lacking a modification in the 3-position (17β-hydroxy, 2-methoxy, and 3H-estradiol) had markedly reduced activity compared with 2ME2 and ENMD-1198 (data not shown). Thus, both a 3-hydroxy (in 2ME2 and ENMD-1269) and a 3-carbamide group (in ENMD-1198 and ENMD-1189) could confer high drug activity.

We also determined the ability of the various analogues to induce apoptosis by measuring nuclear fragmentation and cleavage of the caspase 3 substrate PARP. After 24-hour treatment of cells with 2ME2 analogues at a concentration of 25 μmol/L, marked drug-induced cell rounding and cell death were observed with 2ME2, ENMD-1189, and ENMD-1269, whereas the other analogues did not cause cell toxicity (Fig. 2A; data not shown). When the drugs were added at a low cell density, we also observed marked toxicity with ENMD-1198, consistent with the MTT results above (data not shown). ENMD-1198 has previously been shown to be highly active in different cell lines as well as in vivo (30, 31). The reason and significance for the lack of a proapoptotic effect with ENMD-1198 at higher cell density is currently not clear, but is likely related to drug metabolism. Nuclear fragmentation was apparent with 2ME2, and to a somewhat lesser extent, with ENMD-1189 and ENMD-1269 in both HCT116 and MCF7 cells (Fig. 2B; Table 1). No effect was observed with the other analogues. When measuring drug-induced PARP cleavage after treating the cells for 36 hours with 25 μmol/L of the various analogues, 2ME2 and ENMD-1269 were found to have the strongest effect (Fig. 2C). PARP cleavage induced by ENMD-1189 was somewhat less severe whereas the other analogues did not cause consistent cleavage of PARP. These results indicate that there generally was a good correlation between inhibitory activity of the analogues in the MTT assays and their ability to induce apoptosis, consistent with the proposed mechanism of action of cell cycle arrest due to disruption of microtubule dynamics followed by apoptosis.

2ME2 at high concentration has been reported to depolymerize microtubules by binding to the colchicine binding site of tubulin (3). At low therapeutic concentrations, 2ME2 inhibits microtubule dynamics, which leads to inhibition of the formation of the mitotic spindle apparatus, and consequently, to the arrest of cells in metaphase and induction of apoptosis (4, 5, 7). Therefore, the accumulation of cells in metaphase was measured by determining the mitotic index, measured as the percentage of cells with chromatin condensation. HCT116 cells were treated with 2ME2 analogues for 12 hours, fixed with paraformaldehyde and the nuclei were labeled with 4′,6-diamidino-2-phenylindole. Chromatin condensation was determined by cell counting. ENMD-1269 showed the strongest effect (36% of cells with chromatin condensation compared with 0.5% in the DMSO-treated control), followed by 2ME2 (32%) and ENMD-1189 (20.5%). None of the other compounds induced significant chromatin condensation compared with the control (ENMD-1147, 1.5%; ENMD-1151, 3%; ENMD-1198, 0.5%; ENMD-1506, 1.5%). To rule out that the drug-induced chromatin condensation observed after 12 hours was due to the induction of caspase-dependent apoptosis, cells were cotreated with the pan-caspase inhibitor zVAD-FMK. Very similar results were obtained when zVAD-FMK was present (DMSO, 3%; ENMD-1269, 40%; 2ME2, 34.5%; ENMD-1189, 31%), indicating that chromatin condensation was likely a direct cause of microtubule disruption and cell cycle arrest. The results indicate that there is a very good correlation between analogues that cause chromatin condensation after 12 hours and those that induce apoptosis at 24 and 36 hours. Immunofluorescence staining of HCT116 cells for α-tubulin revealed that similar to paclitaxel (a microtubule-stabilizing drug that served as a positive control), 2ME2 induced the appearance of multipolar spindles (Fig. 2D), as previously reported (7). In contrast, multipolar spindles were not observed with the other 2ME2 analogues (data not shown), suggesting that this is a unique property of 2ME2.

Several studies have reported that 2ME2 induces the production of ROS, specifically of superoxide. The induction of superoxide or its conversion into H₂O₂ has been implicated as the mechanism through which 2ME2 exerts cytotoxicity (11–13). We therefore determined the effect of 2ME2 analogues on cellular superoxide concentrations by measuring the superoxide-dependent oxidation of the fluorescent dye dihydroethidium. Treatment of HCT116 cells with the known superoxide-inducing agents menadione and antimycin A caused a robust increase in dihydroethidium fluorescence (Fig. 3A), thus validating our measurements. 2ME2 at 50 μmol/L induced only a moderate increase in cellular superoxide (39 ± 13% increase compared with control). Cells treated with ENMD-1151 showed the strongest production of superoxide of all 2ME2 analogues (204 ± 54% increase compared with control). ENMD-1189 and ENMD-1269 also induced high levels of superoxide, ENMD-1147 had intermediate effects, and no significant increase in superoxide production was observed in cells treated with ENMD-1198 and ENMD-1506. ENMD-1151 has an amino group in the 2-position, but is otherwise identical to 2ME2, thus suggesting that the 2-methoxy group is not required for superoxide production and may in fact have an inhibitory effect. In contrast, the 2-methoxy group is absolutely required for the antiproliferative and proapoptotic drug effects, given that both 17β-estradiol and ENMD-1151 do not significantly affect cell growth or cell death. The lack of cytotoxic effects with ENMD-1151 also suggested that induction of superoxide is not a major mechanism of action of 2ME2. To confirm this, cells were treated with the antioxidant N-acetylcysteine, which did not exert a protective effect towards 2ME2-induced cell death (Fig. 3B). Cellular superoxide is normally converted into H₂O₂, which exerts many of the toxic effects of superoxide produced in the mitochondria. We used retroviral gene transduction to overexpress cytosolic or mitochondrially targeted catalase which detoxifies cellular H₂O₂. Highly significant overexpression of catalase compared with endogenous protein concentrations was achieved (Fig. 3C). However, this had no effects on 2ME2- and ENMD-1189–induced cell death (Fig. 3D). Taken together, these results suggest that the induction of cellular ROS is not a major mechanism of action for 2ME2 in epithelial cancer cells.

2ME2 and related compounds have been reported to inhibit complex I of the electron transport chain (15, 16). To determine the effects of the 2ME2 analogues on the mitochondrial electron transport chain, oxygen consumption was measured by polarography in mouse liver mitochondria. To allow use of membrane-impermeable NADH as a substrate for complex I, mitochondria were subjected to a freeze-thaw cycle. When using drug concentrations of 50 μmol/L and NADH as the respiratory substrate, treatment with ENMD-1147 was found to cause the greatest inhibition of oxygen consumption (69.2% inhibition; Fig. 4A). 2ME2, ENMD-1151, and ENMD-1189 had intermediate inhibitory effects, whereas only small or nonsignificant effects were observed with ENMD-1198, ENMD-1269, and ENMD-1506.

Given that ENMD-1147 had the strongest inhibitory effect, we further studied the mechanism through which it interacts with the mitochondrial electron transport chain. Electrons that enter the electron transport chain at complex I are transferred to molecular oxygen via respiratory complexes III and IV. Therefore, the inhibitory effect of ENMD-1147 in the presence of NADH as a substrate (Fig. 4B) could be due to inhibition of complex I, III, or IV. When using succinate as a respiratory substrate, succinate-derived electrons are transported via complex II, III, and IV. ENMD-1147 did not inhibit succinate-dependent oxygen consumption (Fig. 4B), ruling out interaction of the drug with complex II, III, or IV and implicating an interaction with complex I as the cause for respiratory inhibition. Consistent with this, inhibition of NADH-dependent oxygen consumption by ENMD-1147 could be reversed upon the addition of succinate (Fig. 4C). The inhibition of complex I by ENMD-1147 was found to be dose-dependent (IC₅₀ of 47 μmol/L; Fig. 4D). Furthermore, the inhibitory effect of ENMD-1147 could be partially reversed by washing the mitochondria in mitochondria isolation buffer after drug treatment. ENMD-1147 at a concentration of 200 μmol/L inhibited mitochondrial oxygen consumption initially by 86.6% compared with the untreated control. After washing, the inhibitory drug effect decreased to 38.6% compared with the control, suggesting reversible binding of the drug to complex I.

ENMD-1147 differs from 2ME2 only in the configuration of the 17-hydroxy group (α-configuration in ENMD-1147 and β-configuration in 2ME2). Thus, our results indicate the unique activity of the 17-α-hydroxy group towards mitochondrial complex I. The compounds bearing a 17β-hydroxy group, 2ME2, ENMD-1189, and ENMD-1151 had intermediate inhibitory effects, whereas the compounds with the 16,17-olefin group had only low activity. As shown above, the α-configuration of the 17-hydroxy group significantly reduced the cytotoxic activity of the compound, indicating that interference with mitochondrial function is not involved in the cytotoxic effects of 2ME2. Inhibition of mitochondrial complex I is therefore unlikely to be an undesired “side effect” of potent proapoptotic 2ME2 analogues. Taken together, our findings indicate that the modification in the 17-position is an important structural determinant of the cellular activity of 2ME2 analogues towards different targets.

Different complex I inhibitors have been shown to induce ROS production originating from electron transport in the mitochondria. We have therefore previously suggested that 2ME2-dependent superoxide production is also due to its interaction with complex I (15). However, comparison of the effects of the different 2ME2 analogues on cellular superoxide production and on complex I activity reveals only a partial correlation. It is possible that upon analogue binding to complex I, inhibition of electron transport and induction of leakage of electrons that reduce molecular oxygen to superoxide is not proportional. Alternatively, it is also possible that other mechanisms contribute to the 2ME2-induced production of cellular superoxide.

To investigate the potency of the various 2ME2 analogues for inhibition of HIF, we first measured hypoxia-induced HIF transcriptional activity using luciferase gene reporter assays (Fig. 5). As expected, incubation of cells under hypoxic conditions (1% oxygen) for 6 hours resulted in a marked increase in reporter activity (14.7-fold increase in MCF7 and 31.6-fold increase in HEK293). Treatment with the various 2ME2 analogues at 25 μmol/L revealed that ENMD-1189 had the strongest inhibitory effect in both cell types (69.4% and 60.8% inhibition in MCF7 and HEK293 cells, respectively). 2ME2 also exhibited clear inhibitory effects (65.3% and 33.9% inhibition in MCF7 and HEK293 cells, respectively), whereas ENMD-1147 and ENMD-1151 had weak effects. No significant inhibition was observed with the other analogues. In control experiments using a luciferase plasmid lacking the HRE, none of the 2ME2 analogues caused a significant decrease in luciferase activity (Fig. 5A and B, right), which indicated that the observed drug effects were not nonspecific due to toxicity.

We next determined the effect of the 2ME2 analogues on hypoxia-induced HIF-1α protein accumulation (Fig. 6). As expected, incubation of MCF-7 and HEK293 cells at 1% oxygen resulted in a marked stabilization of HIF-1α protein. At drug concentrations of 25 μmol/L, only treatment with ENMD-1189 caused a clear decrease in HIF-1α protein accumulation compared with the hypoxia control in both cell lines (45% decrease in MCF-7 cells and 41% in HEK293 cells; Fig. 6A). None of the other analogues showed reproducible effects on HIF-1α.

Our results obtained with ENMD-1189 and 2ME2 indicated stronger effects on HIF transcriptional activity compared with effects on HIF-1α protein accumulation in hypoxia. These results suggest that the drugs may affect HIF transcriptional activity independently of regulating HIF-1α protein abundance. To examine this further, we used a HEK293 cell line that was stably transfected with P402A/P564A mutant HIF-1α under the control of a tetracycline-inducible promoter. Expression of this protein is expected to be independent of posttranscriptional regulation of its protein synthesis, which is likely to be mediated via the HIF-1α 5′- and 3′-untranslated regions, as well as oxygen-dependent regulation of its stability, which is dependent on the hydroxylation of Pro402 and Pro564. Cells were treated with tetracycline in the presence or absence of 2ME2, ENMD-1189, or ENMD-1269. Treatment of cells with tetracycline in normoxia resulted in the induction of the mutant HIF-1α protein and a 2.9-fold increase in HIF transcriptional activity in the absence of 2ME2 analogues (Fig. 6B). Treatment with ENMD-1189 had no effect on mutant HIF-1α protein induction, but reduced HIF transcriptional activity by 37.4%. In contrast, no significant decrease in transcriptional activity was observed with 2ME2 and ENMD-1269, although both drugs caused an increase in mutant HIF-1α protein expression. These results suggest that ENMD-1189, the most potent inhibitor of endogenous HIF-dependent transcription (see Fig. 5), exerts at least part of its effect by directly inhibiting HIF transcriptional activity independently of regulating HIF-1α protein.

Our results show relatively small inhibitory effects of 2ME2 and its analogues on hypoxia-induced HIF-1α protein accumulation and transcriptional activity in MCF7 and HEK293 cells. To determine if higher concentrations of 2ME2 analogues could induce more robust HIF-inhibitory effects, we incubated HCT116 cells under hypoxic conditions for 6 hours in the presence of 25, 50, or 100 μmol/L of the drugs (Fig. 6C). At 25 μmol/L, only ENMD-1198 caused a slight reduction in HIF-1α protein induction. At 50 μmol/L, only ENMD-1198 and 2ME2 had small inhibitory effects, whereas 1 μmol/L of myxothiazol, an inhibitor of the mitochondrial electron transport chain with established potent HIF-inhibitory effects, completely prevented hypoxia-induced HIF-1α protein accumulation (Fig. 6C, top). At 100 μmol/L, ENMD-1198 exhibited consistently very strong inhibition of hypoxia-induced HIF-1α stabilization. ENMD-1269 had intermediate effects, whereas the other analogues caused weak or no inhibition.

A number of studies have reported more potent HIF inhibitory effects, whereas relatively high 2ME2 concentrations were also required in other reports. The differences may be related to cell types and drug exposure times (although we have also only observed small effects upon 24 hours of drug treatment in HCT116 cells). Despite the differences with some previous studies, our results indicate that in the cell types used in this study, considerably higher doses of 2ME2 and its analogues were required to inhibit HIF-1α compared with the concentrations that exerted antiproliferative and proapoptotic effects. Furthermore, there was no good correlation between the antiproliferative and proapoptotic effects and the inhibition of HIF. Thus, ENMD-1269, which potently inhibited cell proliferation and induced apoptosis, was without significant effect on HIF-1α protein and transcriptional activity when used at the same concentration. HIF inhibition is therefore unlikely to contribute to the in vitro cellular activity of 2ME2. However, it remains possible that HIF inhibition occurring in vivo plays a role in the antiangiogenic effects of 2ME2 or ENMD-1198.

With respect to the mechanism of 2ME2-dependent HIF inhibition, it has been previously reported that disruption of microtubules leads to the inhibition of HIF-1α protein synthesis (18). Alternatively, 2ME2-induced production of ROS or inhibition of mitochondrial respiration may lead to the inhibition of HIF-1α protein stabilization in hypoxia (32–36). We did not observe a good correlation of effects on HIF-1α with any of these drug actions. ENMD-1269, which had potent effects on microtubules, ENMD-1151, which was the strongest inducer of superoxide, and ENMD-1147, which had the strongest activity towards mitochondrial complex I, all had no significant effects on hypoxia-induced HIF-1α protein accumulation at their effective drug concentrations. This suggests the existence of an alternative mechanism(s) through which 2ME2 and some of its analogues inhibit the HIF transcription factor. Our data suggest that direct inhibition of HIF transcriptional activity may at least partially be involved.

In summary, our results suggest that targeting of microtubules is the major mechanism of action of 2ME2, whereas the other reported targets of 2ME2, i.e., the induction of cellular ROS, disruption of mitochondrial function, and inhibition of HIF, are unlikely to be important contributors to drug action in epithelial cancer cells. Inhibition of HIF-1α requires supratherapeutic concentrations and does not correlate well with effects on microtubules, raising doubts about both its significance in epithelial cancer cells and the previously reported mechanism of 2ME2-induced HIF-1α inhibition. Our studies also reveal that in addition to the 2-methoxy group, modifications in both the 3- and 17-positions are important determinants of drug activity, and our findings may aid in future drug development.

No potential conflicts of interest were disclosed.

We acknowledge EntreMed, Inc. (Rockville, MD) for providing 2ME2 and analogues; Dr. David Beach (Institute of Cell and Molecular Science, London, United Kingdom) for providing the retroviral expression vector Puro-MaRX; and Dr. Kaye Williams (The University of Manchester, Manchester, United Kingdom) for the pGL3-HRE plasmid.

Grant Support: Singapore National Medical Research Council and National University of Singapore, School of Medicine.

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.