New Soft Alkylating Agents with Enhanced Cytotoxicity against Cancer Cells Resistant to Chemotherapeutics and Hypoxia

A major challenge of modern chemotherapy is the development of drugs that selectively target cancer cells, overcome chemoresistant tumor cells, and have limited toxic effects. To that end, we developed over the past decade a new class of antimitotics called N-phenyl-N′-2-(chloroethyl)ureas (CEU). CEU inhibits the growth of numerous tumors and drug-resistant cell lines (1–5). N-[4-Iodophenyl]-N′-(2-chloroethyl)ureas (CEU-098) and N-[4-(1,1-dimethylethyl)-phenyl]-N′-(2-chloroethyl)urea (CEU-022) were found to block the migration of cancer cells in vitro and display antitumor and antiangiogenic properties in vivo (4, 6). CEUs are nonmutagenic monoalkylating agents that are unreactive toward most cellular nucleophiles, such as glutathione and DNA (2, 7). Protein extracts from cells exposed to [¹⁴C]CEU-022 showed that the drug covalently binds to a limited number of proteins, notably β-tubulin isoform-2 (8). Competition experiments confirmed that CEU-022 and several derivatives bind to the colchicine-binding site and disrupt the microtubule network (8) to block the cell division in G₂/M phase (8). These CEUs are members of a molecular subset that hereafter is designated as βTAC.

Apoptosis is one of the mechanisms by which chemotherapeutic agents induce cancer cell demise. The mitochondrion acts as a convergent point at which the signaling pathways integrate and trigger apoptosis, as shown with various anticancer agents. The mitochondrial-dependent apoptotic pathway results from a mitochondrial membrane permeability transition of the outer mitochondrial membrane culminating in the release of proapoptotic proteins in the cytosol, such as cytochrome c (9). One of the mechanisms leading to membrane permeability transition is the formation of a permeability transition pore (PTP; ref. 10). PTP opening is known to occur in response to different stimuli, such as increased production of reactive oxygen species (ROS), mitochondrial calcium overload, thiol oxidation, adenine nucleotide depletion, and collapse of the mitochondrial membrane potential (ΔΨ_m; ref. 10). Specific contacts between the voltage-dependent anion channel isoform-1 (VDAC-1), adenine nucleotide translocator (ANT), and cyclophilin D are believed to constitute the backbone of PTP (11). Members of the Bcl-2 protein family, which act as checkpoints of apoptosis, play a key role in the membrane permeability transition process (12) and regulate PTP opening/closing in concert with VDAC and ANT (13). Several molecules such as cyclosporin A have been shown to modulate PTP opening and used to decipher its regulation (14).

Chemotherapeutics such as arsenite and lonidamine were found to be direct PTP opening agents (15). For instance, the mitochondrion is recognized as an important target for the development of new anticancer agents (15). Moreover, several lines of evidence suggest that mitochondria of transformed cells exhibit alterations that can be exploited to selectively induce their demise. For example, an increased binding of hexokinase-2 with VDAC-1 is observed in tumor cells, which alleviates survival of cancer cells in hypoxic conditions (16). Inhibition of hexokinase activity by 3-bromopyruvate or decreased binding of hexokinase-2 to VDAC-1 increases the susceptibility of cancer cells to chemotherapeutics (17) and hypoxia (18).

In this work, we report the identification of a subset of cytotoxic CEUs and their corresponding oxazolines that do not alkylate β-tubulin (N-βTAC). N-βTACs have an increased cytotoxicity on hypoxia-resistant fibrosarcoma HT1080 cells generated by mitochondrial DNA depletion. VDAC-1 was identified as a target of N-βTAC.

Reagents and chemicals. Nonradioactive CEU and oxazoline derivatives (Table 1) were kindly provided by Dr. Jean Rousseau from IMOTEP, Inc. (Québec, Canada). The synthesis of [¹⁴C]CEU-025 and [¹⁴C]CEU-027 was carried out as previously described (19). All other drugs and reagents were purchased from Sigma (St. Louis, MO). All drugs assessed in this study were dissolved in DMSO and used at a final concentration of <0.19% (v/v). The human Bcl-2 recombinant protein used (Sigma) is a 68-kDa fusion protein composed of a maltose binding protein moiety followed by a histidine tag and a Bcl-2 protein lacking 21 amino acid residues at the COOH terminus.

Cell lines. Human colon carcinoma HT29, breast carcinoma MDA-MB-231, human fibrosarcoma HT1080, and murine melanoma M21 cell lines were obtained from the American Type Culture Collection (Manassas, VA) and cultured in DMEM (Hyclone, Road Logan, UT) supplemented with 5% bovine calf serum. Wild-type (wt) Chinese hamster ovary (CHO) cells (CHO-10001), colchicine/vinblastine–resistant cells (CHO-VV 3-2), and paclitaxel-resistant cells (CHO-TAX 5-6) were generously provided by Dr. Fernando Cabral (University of Texas Medical School, Houston, TX) and cultured as previously described (8). HT1080 cells partially depleted of mitochondrial DNA were obtained by maintaining wt HT1080 cells for 4 weeks in the presence of 100 ng/mL ethidium bromide in the culture medium supplemented with 300 μg/mL uridine and 2 mmol/L Na pyruvate.

Growth inhibition and cytotoxicity assays. The sulforhodamine B growth assay was done as previously reported (4). Cell cytotoxicity was assessed by the resazurin assay. Cells were seeded onto 96-well microtiter plates and, 24 h later, exposed for 16 or 48 h to the drugs before the addition of medium containing resazurin (25 μg/mL final) for 1 h at 37°C; then, fluorescence measurements [excitation: 530 (±35) nm, emission: 590 (±25) nm] were done with a microtiter plate fluorescence reader (Bio-Tek FL600, Bio-Tek, Winooski, VT). Background fluorescence (BGF) emitted from the control wells containing medium and resazurin without cells was subtracted from fluorescence values obtained in the presence of cells. The percentage of cell viability was calculated as follows: % viability = (mean fluorescence with drug − BGF) / (mean fluorescence without drug − BGF) × 100.

Alkylating potency of CEU. The alkylating potency of CEU and oxazoline versus chlorambucil was determined with the colorimetric assay previously described (8). The alkylation potency P′ refers to the percent kinetics ability of CEU or oxazoline to alkylate 4-(4-nitrobenzyl)pyridine compared with chlorambucil.

Immunocytochemical analysis of microtubule network integrity. HT1080 cells (8 × 10⁵) were seeded onto fibronectin 16 h before exposure to the drugs. Cells were washed with PBS and fixed in 3.7% paraformaldehyde-PBS for 10 min. After two washes with PBS, fixed cells were permeabilized with 0.1% Triton X-100 for 3 min, washed, then blocked for 30 min (10% goat serum) before β-tubulin antibody (1:200) incubation for 1 h at room temperature. After washes, cells were incubated with 1:2,000 dilution of antimouse immunoglobulin G Alexa 488 (Molecular Probes, Eugene, OR) for 45 min at room temperature. All washes were done in PBS containing 10% goat serum or in PBS alone. Cells were then mounted on a microscope slide overnight with slow fade reagent (DakoCytomation, Carpinteria, CA) before analysis under a confocal microscope Eclipse E800 (Nikon, Tokyo, Japan).

Evaluation of mitochondrial cytochrome c release. Cytochrome c release was evaluated by immunocytochemistry and confocal microscopy as described for β-tubulin analysis (see above) except that a cytochrome c antibody was used (1:1,000 dilution; clone no. 6H2.B4, BD PharMingen, San-Diego, CA).

Flow cytometric analysis. Following treatments with oxazoline-195 (OXA-195) or CEU-236 for different periods of time or with DMSO for 24 h, HT1080 cell pellets (2.5 × 10⁵) were resuspended in DMEM containing 5 μmol/L hydroethidine and 40 nmol/L DiOC₆ to measure ROS production and ΔΨ_m, respectively, as described elsewhere (20). After treatment of the cells with the dyes (30 min/37°C), the cells were placed on ice until analysis. DNA content cell cycle analysis was done by propidium iodide staining (8).

One-dimensional and isoelectric focusing/two-dimensional SDS-PAGE analysis. MDA-MB-231 and HT1080 cells were exposed to different concentrations of radioactive (or not) CEU or oxazoline. After incubation, adherent and floating cells were pooled. The cells were pelleted and washed with PBS before their resuspension in 1× Laemmli buffer containing 5% β-mercaptoethanol. The lysates were sonicated for 5 s and boiled for 5 min before one-dimensional electrophoresis. Isoelectric focusing was done with the Protean IEF Cell apparatus according to the instructions of the manufacturer (Bio-Rad, Hercules, CA). Detection of native and alkylated β-tubulin isoforms is based on a one-dimensional electrophoretic shift assay (8). The detection of the native and alkylated forms of VDAC-1 was done with a 1:5,000 dilution of monoclonal anti–VDAC-1 [clone 89-173/033 (ab3) and 89-173/045 (ab4), Calbiochem, San Diego, CA]. All incubations with antibodies were done at room temperature for 2 to 3 h in TBS, 0.1% Tween 20 with 1% milk or in TBS, 0.1% Tween 20 + 5% bovine serum albumin. Poly(ADP-ribose) polymerase (PARP), caspase-3, caspase-8 (clone no. 1C12), and caspase-9 antibodies were obtained from Cell Signaling, Inc. (Beverly, MA). Cytochrome oxidase subunit II (clone no. 12C4) and cytochrome oxidase subunit IV (clone no. 10E8) antibodies were obtained from Molecular Probes.

Mitochondrial isolation. Rat liver mitochondria were purified by differential centrifugation and Percoll gradient according to the procedure reported by Almeida and Medina (21).

Mitochondrial swelling assay. Swelling of rat liver mitochondria was measured with a spectrophotometer (Shimadzu, Kyoto, Japan) at 540 nm. The mitochondrial suspension consisting of rat liver mitochondria (0.5 mg/mL) was incubated in the swelling buffer [70 mmol/L sucrose, 214 mmol/L mannitol, 5 mmol/L HEPES (pH 7.4), 0.5 mmol/L NaPO₄ (pH 7.4)] with or without 5 mmol/L glutamate and 2.5 mmol/L sodium malate. The swelling assays were done at 25°C. Signal decrease at 540 nm is indicative of mitochondria swelling.

VDAC-1 purification. Transmembrane protein purification was done as described by de Pinto et al. (22) with minor modifications. Briefly, cultured cells exposed to CEU-025 were resuspended in PB2 buffer [3% Triton X-100, 10 mmol/L Tris (pH 7), 1 mmol/L EDTA, and 1 tablet of Complete mix of protease inhibitors (Roche, Penzberg, Germany) per 10 mL of buffer]. The protein concentration of the extract was adjusted to 5 mg/mL before applying onto a dry hydroxyapatite/celite (2:1 w/v) column (0.1 g/mg protein). Two milliliters of PB2 buffer were used for elution of VDAC-1.

VDAC-1 oxidation assay. The capacity of CEU and oxazoline to alkylate the cysteinyl residues on VDAC-1 was verified using diamide-mediated oxidation, described elsewhere (23). Total cell or rat liver mitochondria extracts, exposed or not to CEU in Triton 3% buffer, were denatured with SDS 1% (10 min/25°C). All denatured samples were then treated for 30 min with 250 μmol/L diamide. Subsequently, they were boiled in the presence or absence of β-mercaptoethanol (5%) for SDS-PAGE analysis.

Comparison of CEU growth inhibition versus alkylating potency. To optimize the antiproliferative properties of our prototypical CEU-022 (1, 2, 4–6), several CEUs were prepared by substituting the alkyl group at the third or fourth position of the phenyl ring (Table 1). We selected from a CEU library a subset of 9 molecules displaying GI₅₀ ranging from 100 nmol/L to 100 μmol/L and included CEU-091 as a negative control for protein alkylation. CEU-091 has no chlorine atom and is devoid of both alkylating and growth inhibition properties on solid tumors cells (4). CEU-107(R) bears a methyl group (R-enantiomer) on the ethyl linker arm between the urea moiety and the chlorine atom. Interestingly, the S-enantiomer has no growth inhibition activity while having the same electrophilicity (24), thus illustrating the importance of structure-activity relationships in the design of novel CEUs. OXA-152 and OXA-195 are “bioisosteres” of CEU-025 and CEU-027, respectively. OXA-152 and OXA-195 are the cyclized byproducts of the 2-chloroethylamino moiety of CEU into a 2-oxazoline heterocycle. As shown in Table 1, these oxazolines exhibit electrophilic properties and antiproliferative activities similar to or even higher than those of their respective 2-chloroethylurea counterparts. The 4-iodo derivative CEU-098 is a bioisostere of the 4-tert-butyl moiety designed to improve the metabolic resistance of CEU-022 (6). Finally, CEU-236, which is a 3-(5-hydroxypentyl) substituted form of CEU, exhibits the highest cell growth inhibition capacity on several tumor cell lines (7, 25).

N-βTAC: new CEUs and oxazolines that inhibit cancer cell growth without alkylating the β-tubulin colchicine-binding site. Several CEUs were earlier shown to disrupt the cytoskeleton through their covalent binding to the colchicinoid-binding site of β-tubulin isoform-2 (8). As shown in Fig. 1A, CEU-022, CEU-098, CEU-107R, and CEU-236 covalently bind to β-tubulin, which leads to the formation of a protein byproduct exhibiting a faster electrophoretic mobility shift than the native β-tubulin (8). CEU-091 is, as expected, neither cytotoxic nor electrophilic (Table 1). Figure 1A also shows that the electrophoresis of proteins extracted from cells treated with CEU-025, CEU-027, OXA-152, or OXA-195 at concentrations higher than their IG₅₀ did not reveal the formation of any β-tubulin byproducts, suggesting that N-βTAC does not alkylate β-tubulin. In support of this result, protein extracts from MDA-MB-231 cells exposed to [¹⁴C]CEU-022, [¹⁴C]CEU-025, or [¹⁴C]CEU-027 revealed (Fig. 1B) a 50-kDa signal corresponding to the alkylated β-tubulin byproduct, restricted to[¹⁴C]CEU-022 labeling. However, N-βTAC still might partly act on β-tubulin as reversible instead of irreversible antagonists of the colchicine-binding site. To address this issue, we did competition experiments to determine the capacity of N-βTAC to compete with CEU-022–mediated β-tubulin alkylation. Colchicine competes with βTAC for the colchicine-binding site (8), as shown in Fig. 1C. In contrast, neither vinblastine nor any of the N-βTACs tested diminished significantly the formation of the CEU-022 β-tubulin byproduct (Fig. 1C), thus confirming that CEU-025, CEU-027, and their corresponding oxazoline bioisosteres are not reversible antagonists of the β-tubulin colchicine-binding site.

These observations suggest that N-βTAC does not inhibit cell growth as βTAC or classic β-tubulin targeting agents such as colchicinoids and taxoids. To confirm that hypothesis, we did growth inhibition assays using CHO-10001 (wt) and mutant CHO-VV 3-2 and CHO-TAX 5-6 cell lines, which exhibit differential sensitivity to anti–β-tubulin agents through β-tubulin mutations (26, 27). CHO-VV 3-2 cells are resistant to molecules such as vinblastine and colchicinoids and hypersensitive to taxoids such as paclitaxel. Conversely, CHO-TAX 5-6 cells are resistant to paclitaxel and hypersensitive to vinblastine and colchicinoids. CHO-VV 3-2 and CHO-TAX 5-6 were previously found to be resistant and hypersensitive, respectively, to βTAC such as CEU-022 (8, 24) and as shown in this study (Fig. 1D). In contrast, wt and β-tubulin–mutant cell lines displayed similar growth inhibition sensitivities in response to CEU-025, CEU-027, OXA-152, and OXA-195. These results suggest that N-βTAC agents inhibit tumor cell growth through mechanisms unrelated to the microtubule dynamic perturbations triggered by classic microtubule-targeting agents such as colchicine and vinblastine and molecules belonging to the βTAC subset.

βTAC and N-βTAC induce microtubule depolymerization. To assess whether N-βTAC can lead to the microtubule network depolymerization, β-tubulin immunofluorescence analysis was done with confocal microscopy following treatment of HT1080 cells with N-βTAC. CEU-236 was used as a positive control to induce microtubule depolymerization and was compared with OXA-195. Unexpectedly, we found that OXA-195, similarly to CEU-236, induced an early (2 h) massive depolymerization of microtubule (Fig. 2A). This suggests that CEU or OXA-mediated β-tubulin alkylation is not a prerequisite to induce microtubule depolymerization. Figure 2B shows that other N-βTACs (e.g., CEU-025 and CEU-152) provoked also substantial microtubule depolymerization whereas CEU-027 induced weak microtubule depolymerization in HT1080 cells. Together, this supports that βTAC and N-βTAC interfere with the polymerization process of microtubule to different extents and mechanisms.

βTAC and N-βTAC induce apoptosis. We next investigated whether βTAC and N-βTAC induce cell death through apoptosis. We have shown that CEUs and oxazolines, regardless of their β-tubulin alkylating potency, induced PARP cleavage in MDA-MB-231 cells (Fig. 3A). In addition, we have shown that DNA fragmentation occurs following exposure to any of the CEU and oxazoline used in the present study (data not shown). Kinetic analysis of caspase activation in HT1080 cells shows that OXA-195 (50 μmol/L) triggers PARP cleavage after 8 h (Fig. 3B) and the activation of initiators caspase-8 and caspase-9 (Fig. 3B), suggesting that both the intrinsic and extrinsic apoptotic pathways are activated in response to OXA-195. As expected, the effector caspase-3 was activated subsequently to the activation of caspase-8 and caspase-9 (Fig. 3B). Accordingly, cytochrome c is released from mitochondria following exposure to OXA-195 (Fig. 3C) and occurs simultaneously with the induction of caspase-9 activation in response to OXA-195 (e.g., after 8 h; data not shown). Moreover, we observed that other βTACs and N-βTACs induced also the activation of caspase-3, caspase-8, and caspase-9 and the release of cytochrome c from mitochondria in HT1080 and MDA-MB-231 cells (data not shown). Altogether, these observations indicate that CEU and oxazoline, regardless of their capacity to alkylate β-tubulin, induce apoptosis in cancer cells.

βTAC and N-βTAC impair the mitochondrial respiratory chain. ROS production and loss of ΔΨ_m are common cell death manifestations occurring during the early phases of apoptosis (28). We assessed the loss of ΔΨ_m and ROS production changes in response to OXA-195 by flow cytometry using the cationic DiOC₆ and hydroethidine probes, respectively. HT1080 exposed to OXA-195 exhibited an increased percentage of the cell population showing time-dependent ΔΨ_m loss and increased ROS production, both appearing as early as 2 h posttreatment (Fig. 3D). Other βTACs and N-βTACs did also induce ΔΨ_m loss (Supplementary data). However, in the same conditions, these agents triggered contrasting ROS production responses (Fig. 4A, ρ⁺). For example, OXA-152 and OXA-195 induced substantial increase of ROS production, whereas CEU-022 and CEU-025 triggered a decrease of ROS production compared with DMSO (Fig. 4A, ρ⁺). We hypothesized that the abilities of CEUs and oxazolines to modulate ROS production originate from the impairment of the mitochondrial respiratory chain (MRC). To investigate that possibility, we compared the ROS production induced by different CEUs and oxazolines in wt HT1080 cells (ρ⁺) and HT1080 cells that are depleted of mitochondrial DNA (ρ⁰). ρ⁰ cells do not express a number of essential components of the MRC, hence abrogating the production of ROS from MRC (29). Mitochondrial depletion was confirmed by Western blot analysis of cytochrome oxidase subunit II, a mitochondrial DNA–encoded protein (data not shown). Figure 4A shows that mitochondrial DNA depletion abrogated the modulating action of CEU or oxazoline tested on the ROS production level, thus supporting our hypothesis. Moreover, in the presence of the respiration uncoupler carbonyl cyanide m-chlorophenylhydrazone, an agent stimulating the electron flow of MRC, all CEUs and oxazolines increased ROS production in ρ⁺ cells but not in ρ⁰ cells (Fig. 4A). No significant loss of viability was observed at all time periods of treatment used in these experiments as revealed by the forward and side scatter analysis of flow cytometry data (data not shown), thus excluding that the ROS and ΔΨ_m responses are resulting from extensive cell damage. Altogether, these results suggest that CEU and oxazoline interfere with the MRC early in cell death cascade induced by these compounds.

Mitochondrial DNA depletion in HT1080 cells induces opposite effects on the cytotoxic activities of βTAC versus N-βTAC. Interestingly, ρ⁰ cells are relatively resistant to βTAC (Fig. 4B) and classic chemotherapeutic agents such as 5-fluorouracil (5-FU), daunorubicin, chlorambucil, paclitaxel, and vinblastine (Fig. 4C). In contrast, ρ⁰ cells show increase sensitivity to N-βTAC (Fig. 4B). ρ⁰ cells are also much more sensitive to 2-deoxy-d-glucose compared with ρ⁺ cells (Fig. 4C), thus illustrating the glycolytic metabolic dependence of these cells.

OXA-195 induces massive swelling of mitochondria. Our finding that βTAC and N-βTAC induce early MRC impairment led us to hypothesize that these compounds may target the mitochondrion to initiate apoptosis. We speculated that CEU and oxazoline might be PTP opening agents. To investigate this, we studied the capacity of CEU and oxazoline to induce swelling of isolated rat liver mitochondria. Ca²⁺ mitochondrial overloading by calcium chloride (CaCl₂) exposure was used as a positive control inducing mitochondrial swelling. In our assays, swelling of rat liver mitochondria was optimal when using 150 μmol/L Ca²⁺ (Fig. 5A). None of the CEUs and oxazolines (Table 1) assessed with a relatively high concentration (100 μmol/L) induced swelling, except for OXA-195 (Fig. 5A). Under these conditions, OXA-195 induced rapid swelling of mitochondria that caused larger amplitude changes than Ca²⁺ (Fig. 5A). OXA-195–mediated mitochondrial swelling was also observed on mitochondria of intact MDA-MB-231 cells, as confirmed by electron microscope analysis (Supplementary data).

Classical PTP-regulating agents differentially affect the OXA-195– and Ca²⁺-induced swelling. To understand the potential role of PTP-associated proteins in the OXA-195–mediated swelling effect, we have evaluated the effect of inhibitors known to interfere with the function of these proteins. Cyclosporin A, a classic PTP blocker that interferes with the function of cyclophilin D into PTP opening (14), blocked the swelling induced by Ca²⁺ (Fig. 5B). However, it had a weak inhibitory activity on the swelling induced by OXA-195 (Fig. 5B). Bongkrekic acid is an antiapoptotic agent inhibiting PTP opening and converting ANT into its “closed” m-conformation (i.e., exposing the adenine-binding site of ANT toward the matrix; ref. 30). Bongkrekic acid pretreatment (25 μmol/L) considerably inhibited the Ca²⁺-mediated swelling effect (Fig. 5B). Bongkrekic acid could also inhibit OXA-195–induced swelling but only with higher concentrations (50–100 μmol/L; Fig. 5B). Interestingly, cotreatment with cyclosporin A (1 μmol/L) and bongkrekic acid (25 μmol/L) completely blocked the swelling induced by Ca²⁺, but had no such additive effect on the swelling induced by OXA-195 even at high concentrations (5 and 100 μmol/L, respectively; Fig. 5B), as compared with each agent used separately at the same concentrations (data not shown). Atractyloside is also an ANT inhibitor but, in contrast to bongkrekic acid, is a proapoptotic agent that opens PTP. Atractyloside induces the c-conformation of ANT (i.e., exposing the adenine-binding site of ANT toward the cytoplasm; ref. 30). In agreement with other reports (31), we showed that atractyloside itself could induce the swelling of rat liver mitochondria (Fig. 5C). In our assays, this effect was observed at concentrations >2 mmol/L (Fig. 5C). To evaluate the contribution of atractyloside on other swelling inducers, we used 1 mmol/L atractyloside, which per se induced little swelling (Fig. 5C). Pretreatment of rat liver mitochondria with atractyloside slightly enhanced the kinetics of Ca²⁺-mediated swelling (Fig. 5C). Surprisingly, atractyloside pretreatment significantly inhibited OXA-195–mediated rat liver mitochondria swelling (Fig. 5C). Markedly, supplementation of rat liver mitochondria with recombinant Bcl-2 (10 μg/mL) almost completely abrogated Ca²⁺-mediated swelling (Fig. 5D). In contrast, a similar Bcl-2 addition did not prevent the swelling induced by OXA-195. Altogether, these results suggest that OXA-195 can induce mitochondrial swelling via a nonclassic PTP-dependent mechanism.

PTP-regulating agents modulate CEU- and oxazoline-mediated growth inhibition responses. We next investigated the effect of PTP inhibitors on the growth inhibition responses induced by CEU or oxazoline. Unexpectedly, pretreatment of HT1080 cells with the PTP blocker cyclosporin A increased the growth inhibition responses induced by N-βTAC and βTAC (Supplementary data). In contrast, pretreatment of cells with atractyloside substantially reduced the antiproliferative action of N-βTAC but not of βTAC (Supplementary data).

CEU and oxazoline bind covalently to the VDAC-1. Previous experiments conducted by Legault et al. (8) suggested that CEU and oxazoline derivatives can covalently bind to proteins that are unrelated to β-tubulin to induce their growth inhibition and proapoptotic actions. SDS-PAGE autoradiograms of cells treated with [¹⁴C]CEU-022, [¹⁴C]CEU-025, or [¹⁴C]CEU-027 (Fig. 1B) revealed the radiolabeling of several proteins, notably a 34-kDa (p34) protein. The identification of p34 was done using isoelectric focusing/two-dimensional electrophoresis of proteins extracted from MDA-MB-231 cells treated with [¹⁴C]CEU-025. Figure 6A shows the presence of the [¹⁴C]CEU-025 labeled p34 byproduct exhibiting an estimated isoelectric point of ∼8. p34 was localized by silver staining on a corresponding two-dimensional gel, excised from the gel, trypsinized, and then analyzed by mass spectrometry. It allowed the identification of the voltage-dependence anion channel isoform-1 (VDAC-1) as the most likely target of CEU-025 (data not shown). Western blot analysis with anti–VDAC-1 revealed a spot that perfectly matches the radioactive spot onto the nitrocellulose membrane (Fig. 6A), strongly suggesting that VDAC-1 is a target of [¹⁴C]CEU-025. Similar results were observed for protein extracts labeled with [¹⁴C]CEU-022 and [¹⁴C]CEU-027 (data not shown). Furthermore, purified VDAC-1 isoform isolated from MDA-MB-231 cells treated with [¹⁴C]CEU-025 also displayed corresponding anti–VDAC-1 and radioactive signals (Fig. 6B).

CEU-025 alkylates cysteine residue(s) of VDAC-1. The human VDAC-1 isoform possesses only two conserved cysteinyl residues (Cys¹²⁷ and Cys²³²) able to form a disulfide bridge via a redox reaction, which is impaired by alkylating agents (32). We examined the ability of CEU and oxazoline to block the diamide-mediated disulfide bridge formation. Under nonreducing conditions, the disulfide bridge of VDAC-1 (oxidized form) generated by the addition of diamide exhibits an increased electrophoretic mobility of VDAC-1 easily distinguishable from the fully reduced form, as described by De Pinto et al. (32) and shown in Fig. 6C. The monothiol alkylating agent N-ethylmaleimide abolished the migration shift of VDAC-1 induced by diamide (data not shown). Figure 6D shows that increasing the time of exposure of MDA-MB-231 cells to CEU-025 leads to a progressive formation of the reduced band of VDAC-1 in the presence of diamide. Similar results were obtained in response to CEU-O22 and CEU-027 (data not shown). These results strongly suggest that CEU-022, CEU-025, and CEU-027 alkylate VDAC-1 on at least one of the two cysteinyl residues.

The main objective of this work was to investigate the mechanism underlying the cytotoxic and growth inhibition properties of βTAC and a newly discovered subset of CEUs and oxazolines called N-βTAC. Regardless of their potency to bind to β-tubulin, we found that CEU and oxazoline induced apoptosis in cancer cells. The activation of initiator caspase-8 and caspase-9 in response to OXA-195 supports the view that the intrinsic and extrinsic apoptotic pathways can be activated by βTAC and N-βTAC.

An important finding is that the cytotoxicity of N-βTAC was increased in ρ⁰ cells when compared with ρ⁺ cells. Depletion of mitochondrial DNA reportedly increases the resistance of cancer cells toward chemotherapeutic drugs (33). Accordingly, we observed that HT1080 ρ⁰ cells are less sensitive than ρ⁺ cells to various antineoplastic agents, such as microtubule-targeting agents paclitaxel and vinblastine. The mechanisms responsible for such resistance are not fully understood. Mitochondrial DNA depletion may be involved in the etiology of cancer, notably by increasing the expression of genes that enhance glycolysis, block apoptosis, and increase the invasive behavior of cancer cells (34). These genetic changes might be essential for the survival of cancer cell clusters in tumor hypoxic environments. Accordingly, ρ⁰ cells are much more sensitive than ρ⁺ cells to the glycolytic inhibitor 2-deoxy-d-glucose. Glycolytic inhibitors constitute a new chemotherapeutic strategy to target hypoxia-resistant tumor cells. Although we do not know yet whether and how N-βTAC affects glycolysis, our study suggests that these agents should be tested for their capacity to selectively target hypoxic chemoresistant tumor cells.

Based on time course experiments, both βTAC and N-βTAC were found to induce early mitochondrial perturbations such as collapse of ΔΨ_m and MRC-dependent ROS production, their onset (2 h) preceding the apparent release of cytochrome c (8 h, data not shown). The roles of ΔΨ_m loss and ROS production in apoptosis have been the subject of controversies, but both were suggested to be initial events inducing membrane permeability transition and cytochrome c release (28). Our observations suggest that βTAC and N-βTAC stimulate ROS production through impairment of the MRC, but the contribution of those mitochondrial perturbations to the cytotoxic mechanism and the targets of these compounds remain to be determined. Mitochondrial ROS stimulation by prooxidants was shown to induce PTP via an ANT dimerization–dependent process (35). Increased ROS production can induce lipid peroxidation and loss of mitochondrial lipid cardiolipin, enabling cytochrome c detachment from cardiolipin (36) and altering membrane fluidity (37). Those alterations may in turn uncouple MRC and further enhance ROS production. In this context, it is noteworthy that some CEUs were found to interact with lipids and alter membrane fluidity (38, 39).

OXA-195 was found to trigger a rapid swelling of rat liver mitochondria, in contrast to other CEUs and oxazolines tested, involving an atypical mechanism, which is notably insensitive to or poorly affected by the PTP inhibitor cyclosporin A and Bcl-2 supplementation. Intriguingly, the PTP opening agent atractyloside diminished the extent of the rat liver mitochondria swelling and the growth inhibition induced by some N-βTAC. In contrast, cyclosporin A enhanced the growth inhibition induce by βTAC and N-βTAC. It is unclear how CEU and oxazoline induce these pharmacologically dependent responses. Yet, it further emphasizes that βTAC and N-βTAC induce different cytotoxic mechanisms and suggests that some of these compounds, such as OXA-195, induce an unusual PTP-dependent apoptosis process. Accordingly, we found that OXA-195 triggers cytochrome c release from isolated rat liver mitochondria (data not shown). However, mitochondrial swelling might not be essential to initiate membrane permeability transition in response to CEU or oxazoline because CEU-025, which does not induce swelling, induced cytochrome c release (data not shown). Mitochondrial swelling before rupture of outer mitochondrial membrane is not a prerequisite for cytochrome c release (31).

We found that VDAC-1 is a potential target of N-βTAC (CEU-025 and CEU-027). VDAC-1 is a channel-forming protein present in the outer mitochondrial membrane, which allows the exchange of most metabolites, notably ADP/ATP between the cytosol and the internal compartments of the mitochondrion (40). VDAC-1 was proposed to be an intrinsic PTP component that by itself, or in concert with other proteins such as ANT and proteins of the proapoptotic Bcl-2 family, constitutes a channel allowing the passage of cytochrome c (13). In that context, the conductance properties of VDAC-1 were shown to be important whereas closed conductance states of VDAC-1 stimulate membrane permeability transition. Bcl-2 protein members modulate the conductance state of VDAC-1 and thus control the onset of membrane permeability transition (41). Interestingly, cytochrome c release, induced by ROS, was found to be VDAC-1 dependent and cyclosporin A independent (42). The binding of hexokinase-1 or hexokinase-2 to mitochondria was shown to involve interactions with VDAC-1 (17, 43). Interestingly, these interactions increase the glycolysis-dependent metabolism of cancer cells and prevent the induction of the apoptosis induced by different apoptotic stimuli such as staurosporine (44–47). In the near future, it will be important to assess whether N-βTAC–dependent VDAC-1 alkylation affects such interactions with hexokinase-1 and/or hexokinase-2 to selectively kill these hypoxia-resistant cells. It should be pointed out that the apparent ability of βTAC compound (CEU-022) to bind to VDAC-1 raises issues in this context. Nevertheless, the ability of radiolabeled βTAC and N-βTAC to alkylate various proteins suggests that the cell growth inhibition mechanisms triggered by these compounds are complex and do not rely per se on the selective and specific alkylation of a single protein. In addition, the alkylation of a protein or a subset of proteins may modulate or be modulated by the alkylation of other proteins. For example, the covalent binding of βTAC to β-tubulin could modify the responses of VDAC-1 alkylation and vice versa. Thus, despite their inability to bind to β-tubulin, N-βTAC could destabilize the microtubule network through such alkylation interplays. In this context, it is known that mitochondria are dynamically associated with microtubule and that VDAC-1 is a binding site for β-tubulin (48). VDAC-1 was recently reported to be required for Bim-dependent apoptosis (49), a Bcl-2 proapoptotic member released from microtubule on exposure to some cell death stimuli (50).

In conclusion, we have identified a new class of soft alkylating agents, called N-βTACs, which induce apoptosis and microtubule depolymerization of cancer cells. In contrast to βTAC, N-βTAC does not covalently bind to β-tubulin and shows an increased cytotoxicity against cells exhibiting mitochondrial DNA depletion, which leads to resistance toward other chemotherapeutics. N-βTAC molecules are promising tools for the design of new antineoplastic agents that target chemoresistant or hypoxia-resistant cancer cells. These molecules may contribute to an understanding of the signaling pathways underlying such resistance mechanisms. ANT and VDAC-1 might be key players in controlling PTP-dependent apoptosis, microtubule-mitochondrion interactions, and energetic balance either as protein binding partners or an independent tandem with specific roles.

Grant support: Cancer Research Society of Canada.

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.

We thank Dr. M.R. Ven Murthy and Marc-Édouard Mirault for critical reading of the manuscript; Dr. Fernando Cabral for β-tubulin mutant CHO cells; Dr. Maurice Dufour for helping with the fluorescence-activated cell sorting analysis; and Claude Marquis for excellent technical assistance.