Microtubule Binding and Disruption and Induction of Premature Senescence by Disorazole C₁

Abstract

Disorazoles comprise a family of 29 macrocyclic polyketides isolated from the fermentation broth of the myxobacterium Sorangium cellulosum. The major fermentation product, disorazole A₁, was found previously to irreversibly bind to tubulin and to have potent cytotoxic activity against tumor cells, possibly because of its highly electrophilic epoxide moiety. To test this hypothesis, we synthesized the epoxide-free disorazole C₁ and found it retained potent antiproliferative activity against tumor cells, causing prominent G₂/M phase arrest and inhibition of in vitro tubulin polymerization. Furthermore, disorazole C₁ produced disorganized microtubules at interphase, misaligned chromosomes during mitosis, apoptosis, and premature senescence in the surviving cell populations. Using a tubulin polymerization assay, we found disorazole C₁ inhibited purified bovine tubulin polymerization, with an IC₅₀ of 11.8 ± 0.4 μM, and inhibited [³H]vinblastine binding noncompetitively, with a K_i of 4.5 ± 0.6 μM. We also found noncompetitive inhibition of [³H]dolastatin 10 binding by disorazole C₁, with a K_i of 10.6 ± 1.5 μM, indicating that disorazole C₁ bound tubulin uniquely among known antimitotic agents. Disorazole C₁ could be a valuable chemical probe for studying the process of mitotic spindle disruption and its relationship to premature senescence.

Natural products have provided a plethora of pharmacologically useful drugs and chemical probes. The disorazole polyene macrodiolides were first isolated from the myxobacterium Sorangium cellulosum in 1994 and characterized to have significant antifungal activity with no antibacterial activity (Jansen et al., 1994). Initial biochemical and pharmacological studies were restricted to the major fermentation product, disorazole A₁ (Fig. 1), which blocks cell proliferation, causes G₂/M phase arrest and loss of microtubules, and induces apoptosis. Moreover, it blocks in vitro polymerization of tubulin (Elnakady et al., 2004; Kopp et al., 2005). Disorazole A₁ contains a highly electrophilic divinyl oxirane moiety that we hypothesized might mediate the mitotic arrest and inhibition of tubulin polymerization through covalent binding to tubulin (Wipf et al., 2006). The highly electrophilic divinyl oxirane of disorazole A₁ is generally not viewed as a therapeutically desirable moiety; therefore, we synthesized the rare family member disorazole C₁ (Fig. 1), which is devoid of reactive groups. It is remarkable that disorazole C₁ retained antimitotic activity (Wipf and Graham, 2004; Wipf et al., 2006). Structural analogs suggested that the functional group array of disorazole C₁ and its three-dimensional conformation were critical for biological activity, but little is known about its mechanism of action. In the current comprehensive report, we demonstrate that disorazole C₁ has potent antiproliferative activity against a wide variety of human tumor cells, disrupts cellular microtubule integrity, blocks tubulin polymerization in vitro, binds tubulin in a unique manner, and causes apoptosis and premature cellular senescence, all attributes associated with a promising anticancer agent.

Fig. 1.

Chemical structures of disorazoles.

Materials and Methods

Cell Culture Reagents and Proliferation Assays. Cells were cultured in the following media supplemented with 10% fetal bovine serum (VWR, West Chester, PA): PtK2 rat kangaroo kidney epithelial cells in minimal essential medium-α; head and neck squamous cell carcinoma cell lines in Dulbecco's modified Eagle's medium; A549 and WI-38 fibroblasts in basal medium Eagle; UPCI:SCC103 in minimal essential medium; and MDA-MB-231, PC-3, and 2008 in RPMI; and HCT116 in McCoy's 5A. Unless otherwise indicated, all media, sera, and supplements were obtained from Invitrogen (Carlsbad, CA), and other reagents were from Sigma-Aldrich (St. Louis, MO).

Inhibition of growth was determined spectrophotometrically with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide or by measuring fluorescence in live cells using an alamar blue-based assay (Promega, Madison, WI). Alternatively, in some studies, we fixed cells in 3.7% formaldehyde and stained nuclei with 2 μg/ml Hoechst 33342 dye to quantify cells using an ArrayScan VTi (Thermo Fisher Scientific, Waltham, MA). The 50% growth-inhibitory concentrations (IC₅₀) of test agents were calculated after a 72-h incubation. For reversibility studies, we treated A549 cells for 1 h with vehicle, 10 nM disorazole C₁, or 1 μM nocodazole. Cells were either continuously exposed to compound in complete medium or briefly (1 h) exposed to a compound, which was removed by washing three times with complete medium. Live cells were counted by trypan blue exclusion and normalized to control cells. For studies with quiescent cell studies, confluent WI-38 cells were treated for 72 h with either DMSO or disorazole C₁.

Immunofluorescence Detection of Microtubules. PtK2 cells (3 × 10⁴ cells/35 mm well) were plated 48 h before treatment. Cells were rinsed twice to remove antibiotics and incubated for 24 h with compounds diluted in antibiotic-free growth medium. Cells were processed to visualize microtubules and DNA, and digital images were collected and manipulated as described previously (Stout et al., 2006). A549 cells were plated on coverslips in six-well plates for 24 h before compound treatment, fixed in 3.7% paraformaldehyde, and incubated in 0.1% Triton X-100 at 4°C for 7 min. A 1% solution of bovine serum albumin in phosphate-buffered saline plus 0.1% Tween was used as the blocking buffer. Microtubules were visualized using an anti-α-tubulin mouse monoclonal antibody (Abcam Inc., Cambridge, MA) with goat-anti-mouse Alexa Fluor488 (Invitrogen) secondary antibody diluted in 1% bovine serum albumin/phosphate-buffered saline plus 0.1% Tween. Nuclei were visualized with the fluorescent dye 4,6-diamidino-2-phenylindole and an Olympus BX60 epifluorescence microscope (Olympus, Tokyo, Japan). Images were taken using a Hamamatsu Argus-20 CCD camera and image processor (Hamamatsu Corporation, Bridgewater, NJ).

Radioligand Binding and Inhibition of Tubulin Assembly. Tubulin was isolated from bovine brain as described previously (Hamel and Lin, 1984). [³H]Vinblastine (specific activity, 61-315 mCi/mmol) and Sephadex G-50 were from GE Healthcare (Chalfont St. Giles, UK). Handee centrifuge columns and the bicinchoninic acid protein assay kit were from Pierce Chemical (Rockford, IL). [³H]Do-lastatin 10 (specific activity, 26-133 mCi/mmol), dolastatin 10, and paclitaxel were from the Drug Synthesis and Chemistry Branch at the National Cancer Institute (Rockville, MD). Radiolabeled ligand binding to tubulin was measured by centrifugal gel filtration and scintillation spectrometry as described previously (Bai et al., 1995b), with minor modifications (Wang et al., 2007). The average stoichiometry of binding in the control reaction mixture was 0.63 mol [³H]vinblastine or [³H]dolastatin 10/mol of tubulin. For the tubulin assembly studies, we used homogeneous bovine brain tubulin (10 μM) and previously described methods (Wang et al., 2007).

Apoptosis and Senescence Assays. A549 cell were analyzed for apoptosis using a commercially available fluorescein terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) method with fluorescence microscopy as described by the manufacturer (Roche In Situ Cell Death Kit; Roche Diagnostics, Indianapolis, IN). For senescence measurements, A549 cells (3.5 × 10⁵ cells/well) were plated 24 h before treatment with DMSO vehicle or test compounds for 1 to 7 days. For 7-day treatments, individual wells were split on day 3 or 4 to avoid the cultures reaching confluence. β-Galactosidase staining was performed as described previously (Dimri et al., 1995) and quantified both by visual counting by a naive observer or by acquiring images with the bright-field module on an ArrayScan VTI and using the Cellomics Compartmental Analysis V3 Bioapplication (Cellomics, Inc., Pittsburgh, PA) to define cells with positive β-galactosidase staining (mean of 20 acquired fields/well). For BrdU incorporation, A549 cells were pretreated in thymidine-free medium, incubated for 6 h with BrdU, and visualized using a cell proliferation kit (GE Healthcare). BrdU images were acquired (20 fields/well) with an ArrayScan VTi and analyzed using the Cellomics Compartmental Analysis V3 Bioapplication1 to determine the percentage of cells with fluorescence greater than the set threshold and to quantify BrdU-positive cells. For Western blotting, A549 cells were treated with compounds for 7 days as stated above followed by cell harvesting in ice-cold lysis buffer (50 mM Tris, pH 7.6, containing 150 mM NaCl, 1 mM EDTA, 0.1% SDS, and 1% Triton X-100) supplemented with protease and phosphatase inhibitors. Protein concentrations in cell lysates were determined with the BCA protein assay kit (Promega). Individual proteins were detected by standard Western blotting methods with the following antibodies: p53, pRb, phospho-Rb (Ser780), γH2AX, and GAPDH antibodies (Cell Signaling Technology Inc., Danvers, MA); p21 antibody (EMD Biosciences, San Diego); vinculin (Santa Cruz Biotechnology, Inc., Santa Cruz, CA); cyclin D antibody (BD Biosciences, San Jose, CA). Discodermolide was obtained from Novartis (Basel, Switzerland), and disorazole A1 was from Æterna Zentaris GmbH (Frankfurt, Germany).

Results

Disorazole C₁ is a Potent Inhibitor of Cell Proliferation, Interferes with Tubulin Polymerization, and Binds to Tubulin. As an extension of our previous study with HeLa cells (Wipf et al., 2006), we examined the growth-inhibitory activity of disorazole C₁ using 11 other human tumor carcinoma cell lines (Table 1). Disorazole C₁ was remarkably potent, with an average 50% growth-inhibitory concentration (IC₅₀) of 1.7 ± 0.6 nM (S.E.M.). Head and neck cancer cell lines were particularly sensitive to disorazole C₁, with an average IC₅₀ value of 358 ± 56 pM. The p53 wildtype HCT116 cells exhibited similar sensitivity to disorazole C₁ compared with the p53 null MDA-MB-231 cells. When quiescent WI-38 fibroblasts were exposed to disorazole C₁ for 72 h, we observed no change in cell number or morphology. In contrast, proliferating WI-38 cells were sensitive to disorazole C₁, indicating that active cell division was important for the cytotoxic effects of disorazole C₁.

TABLE 1.

Growth inhibition with microtubule-disrupting agents

Human cells were plated in 96-well plates, allowed to attach for 24 h, and treated for 72 h with DMSO or a range of concentrations of test agents. Cell number was determined spectrophotometrically at 570 nm minus absorbance at 690 nm after exposure to 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium or by measuring fluorescence (excitation, 560; emission, 590) in live cells using the Cell Titer Blue Assay (Promega).

Cell Line	Description	IC50 ± S.E.M.a
		Disorazole C1	Vincristine	Vinblastine
			nM
A549	Lung carcinoma	2.21 ± 0.23	21.62 ± 2.68	1.52 ± 0.09
PC-3	Prostate adenocarcinoma carcinoma	1.57 ± 0.10	4.68 ± 0.29	0.86 ± 0.08
MDA-MB-231	Breast epithelial adenocarcinoma	3.53 ± 0.19	7.16 ± 0.37	1.34 ± 0.21
2008	Ovarian carcinoma	1.91 ± 0.23	21.81 ± 2.92	2.24 ± 0.16
HCT116	Colorectal carcinoma	1.09 ± 0.41	5.62 ± 0.33	1.40 ± 0.07
UPCI:SCC104	Oral squamous carcinoma	6.87 ± 0.54	2.98 ± 0.22	1.13 ± 0.18
PCI 15A	Head and neck squamous carcinoma	0.26 ± 0.03	0.41 ± 0.04	1.86 ± 0.14
PCI 15B	Head and neck squamous carcinoma	0.34 ± 0.03	0.94 ± 0.20	1.81 ± 0.27
PCI 37A	Head and neck squamous carcinoma	0.53 ± 0.07	0.66 ± 0.06	2.51 ± 0.06
PCI 37B	Head and neck squamous carcinoma	0.26 ± 0.02	0.49 ± 0.05	3.58 ± 0.45
UMSCC22A	Head and neck squamous carcinoma	0.40 ± 0.06	1.45 ± 0.24	3.27 ± 0.70
WI-38 confluent	Fibroblasts	>100	ND	>100
WI-38 proliferating	Fibroblasts	1.74 ± 0.78	8.58 ± 0.78	1.54 ± 0.13

^aIC₅₀ values are mean of eight determinations.

Disorazole A₁ was reported previously to inhibit tubulin polymerization in vitro at substoichiometric concentrations and block cell proliferation at the G₂/M phase (Elnakady et al., 2004). We previously observed that disorazole C₁ caused G₂/M phase arrest with HeLa cells (Wipf et al., 2006). As illustrated in Fig. 2A, disorazole C₁ clearly disrupted GTP-mediated tubulin assembly in a concentration-dependent manner, with an IC₅₀ of 11.8 ± 0.4 μM (Fig. 2A), indicating a compound/tubulin stoichiometry of approximately 1:1.

Fig. 2.

Disorazole C₁ disrupts microtubule assembly and binds to tubulin in vitro. A, bovine brain tubulin was preincubated with disorazole C₁ or vehicle and monosodium glutamate for 15 min at 30°C. The reaction mixture was cooled to 0°C, GTP was added, reaction mixtures were transferred to cuvettes, held at 2.5°C in a spectrophotometer, and absorbance (turbidity) was read at 350 nm. After a baseline was established, the temperature was raised to 30°C (∼1 min). After 20 min, the temperature was returned to 2.5°C. GTP alone (no test compound) was assigned as 100% assembly (positive control) and DMSO alone (no test compound, no GTP) as 0% assembly (negative control). Hanes-Woolf plot of disorazole C₁ inhibition of [³H]vinblastine (B) or [³H]dolastatin 10 (C) binding to 10 μM bovine brain tubulin. The mixtures were incubated for 30 min at room temperature and bound versus free radiolabeled ligand was separated using Sephadex G-50 columns. The amount of radiolabeled ligand bound to tubulin was determined using scintillation spectrometry. The substrate concentration ([S]) is indicated on the abscissa and the ratio of the substrate concentration to the reaction velocity ([v]) indicated on the ordinate.

We next probed the ability of disorazole C₁ to inhibit the binding of [³H]vinblastine or [³H]dolastatin 10 to purified tubulin as described previously (Bai et al., 1995b) (Fig. 2, B and C). As illustrated by the intersecting lines in the Hanes-Woolf plot (Fig. 2B), disorazole C₁ inhibited [³H]vinblastine binding in an noncompetitive manner, with a K_i of 4.5 ± 0.6 μM. With this linear transformation, the lines would be parallel if the inhibition was competitive. Dolastatin 10 binds at the depsipeptide binding site on tubulin, which overlaps with the vinca alkaloid binding site (Bai et al., 1995a,b). It is interesting that we also found noncompetitive inhibition of [³H]dolastatin 10 binding by disorazole C₁, with a K_i of 10.6 ± 1.5 μM, suggesting that disorazole C₁ might bind tubulin uniquely among known antimitotic agents.

We next treated A549 cells with IC₅₀ concentrations of disorazole C₁ or vinblastine and visualized microtubule integrity by immunofluorescence microscopy (Fig. 3). Compared with vehicle control, disorazole C₁ disrupted microtubules as early as 24 h (Fig. 3b) with complete disruption within 72 h (Fig. 3c). Even within 1 h, disorazole C₁ (2 nM, Fig. 3f; 10 nM, Fig. 3g; 20 nM, Fig. 3h) and 20 nM vinblastine (Fig. 3i) caused a dissolution of microtubules.

Fig. 3.

Disorazole C₁ disrupts microtubules in vivo. Microtubules in A549 cells were partially disrupted after disorazole treatment at 2 nM for 24 h (b) and more completely at 72 h (c) compared with DMSO vehicle control (a). The disruption was similar to that seen with 2 nM vinblastine treatment at 24 h (d) and 72 h (e). A concentration response effect on microtubules after a 1-h treatment was observed with 2 nM (f), 10 nM (g), and 20 nM (h) disorazole C₁, similar to 20 nM vinblastine (i). With both compounds, apparent tubulin aggregation was observed at higher concentrations (h and i). Cells were fixed and stained with anti-α-tubulin antibody to visualize microtubules (green) and 4,6-diamidino-2-phenylindole to visualize nuclei (blue).

To further assess the effects of disorazole C₁ on microtubules at interphase and mitosis, we examined PtK2 rat kangaroo kidney epithelial cells, which are widely used to study antimitotics because they have a flat morphology and large chromosomes, allowing greater visualization. PtK2 cells were treated for 24 h in the presence of disorazole C₁ (5, 25, or 100 nM) or the inactive but structurally closely related analog O1OA (Fig. 1). O1OA had no effect on microtubules at interphase or mitosis, where chromosomes aligned normally along the metaphase plate (Fig. 4). In contrast, disorazole C₁ completely disrupted microtubules at both concentrations. At mitosis, complete misalignment of chromosomes was evident, as were multipolar spindles and dispersed chromosomes with no attached microtubules. These results, in combination with the data described in Fig. 3, confirmed that disorazole C₁ treatment had a dramatic effect on cellular microtubules, causing disruption of normal mitotic processes.

Fig. 4.

Disorazole C₁ disrupts microtubules at interphase and metaphase. Mammalian PtK2 cells were treated with the indicated concentrations of either O1OA (inactive analog of disorazole C₁) or disorazole C₁ for 24 h at 37°C. Cells were fixed and stained with anti-α-tubulin antibody to visualize microtubules (red) and Hoechst 33342 to visualize chromosomes (blue). All images were collected with a Nikon E600 microscope (Nikon, Tokyo, Japan) equipped with a Roper CoolSnap HQ digital camera (Roper Scientific, Trenton, NJ) at 40× (interphase) or 100× (mitotic spindles) objectives. Top, interphase; bottom, mitosis. Bars, 10 μm.

We next determined the durability of the growth-inhibitory effects of disorazole C₁. A549 cells were treated for 1 h with 10 nM disorazole C₁, 1 μM nocodazole, or DMSO vehicle and were maintained in medium containing compound (contin) or washed and maintained in medium lacking compound (wash) and incubated for 1 to 144 h (Fig. 5A). The cytotoxic effects of nocodazole, a reversible inhibitor of mitosis, were completely reversed with compound washout. In contrast, cell viability continued to decline during the 72 h in cultures exposed briefly (1 h) or continuously to 10 nM disorazole C₁. The number of surviving cell remained relatively constant between 72 and 144 h, suggesting growth arrest and possible senescence. A microscopic analysis of microtubule integrity also suggested the disorazole C₁-mediated microtubule disruption was not readily reversible. Furthermore, the IC₅₀ values for cells treated for 1 h or 7 days were similar (2.2 versus 1.5 nM). These growth-inhibitory results were consistent with a previous report using disorazole A₁, in which the cytotoxic effects were prolonged (Elnakady et al., 2004). Disorazole C₁ at 10 nM caused some apoptosis in A549 cells (Fig. 5B), although it was much less effective than what has been reported for disorazole A₁ (Elnakady et al., 2004). Thus, treatment of A549 cells with 2 nM disorazole C₁ for 24 or 48 h caused no significant apoptosis as detected by a TUNEL assay, although treatment with 10 nM for 48 h did cause detectable apoptosis (Fig. 5B). These results indicate that the highly reactive vinyl oxirane subunit in disorazole A₁ was dispensable for the prolonged cytotoxicity caused by the disorazole family members.

Fig. 5.

Disorazole C₁ causes irreversible growth inhibition. A549 cells were treated for 1 h with 10 nM disorazole C₁, 1 μM nocodazolem, or DMSO vehicle followed by either washout of compound (wash) or continuous exposure to compound (contin) for 1, 24, 48, 72, and 144 h. Live cells were counted by trypan blue exclusion. The number of viable cells (trypan blue negative) is graphed as the percentage of viable cells per milliliter normalized to the DMSO control. Data are representative of two independent experiments. B, A549 cells were incubated with doxorubicin (50 nM) or disorazole C₁ (2 or 10 nM) for 24 (open bars) or 48 (black bars) h, and apoptosis was determined with a TUNEL assay. n = 3; bars, S.E.M.

Disorazole C₁ Induces Premature Senescence. Because drug-resistant cells have been so valuable in understanding the mechanism of action of new small molecules, we attempted to generate A549 cells that were resistant to disorazole C₁ using chronic exposure to increasing compound concentrations in culture. Although our initial efforts were unsuccessful, when we examined the population that survived a 1-week exposure, we observed cells with an enlarged, flattened morphology. These cells ceased to divide but remained viable for >50 days in the presence or absence of disorazole C₁. We tested whether or not these cells entered premature senescence using a classic β-galactosidase staining procedure at pH 6.0 after treatment of A549 cells for 7 days with 1 to 10 nM disorazole C₁ (Fig. 6). Cells treated with DMSO or the inactive O1OA analog displayed only a few sparse β-galactosidase-positive cells, whereas the positive control, doxorubicin, showed abundant β-galactosidase staining at the IC₅₀ concentration of 50 nM. Cells treated with an IC₅₀ concentration of discodermolide also displayed high levels of β-galactosidase-positive cells, whereas an equipotent growth-inhibitory concentration of vinblastine had no effect consistent with previous results (Chang et al., 1999; Klein et al., 2005). Disorazole C₁ treatment resulted in positive β-galactosidase staining even at concentrations below the growth IC₅₀ (1 nM). The percentage of β-galactosidase-positive cells with disorazole C₁ treatment was comparable with that seen with doxorubicin. Disorazole C₁ also produced positive β-galactosidase human colon cancer (HCT116), lung cancer (H1299), and oral squamous carcinoma (UPCI:SCC103) cells, but not PC-3 prostate cancer cells (data not shown).

Fig. 6.

Disorazole C₁ causes premature senescence. A, A549 cells were treated for 7 days with disorazole C₁, IC₅₀ concentrations of discodermolide (DISCO; 25 nM), vincristine (VINCR; 20 nM), or vinblastine (VBL; 2 nM), followed by β-galactosidase staining, which was quantified using the bright-field module and compartmental bioapplication analysis on the ArrayScan VTI with values representative of at least two independent experiments. Doxorubicin (DOX; 50 nM) was used as a positive control for senescence, whereas DMSO vehicle and the inactive analog O1OA (10 nM) were used as negative controls. B, A549 and HCT116 cells were treated for 7 days, and cell lysates were probed by Western blot analysis of protein markers of senescence. GAPDH and vinculin, which were used as loading controls, are located directly below the examined protein lanes.

Alterations in the protein expression profiles of some of the classic markers of premature senescence also supported the hypothesis that these metabolically active, nondividing cells had undergone premature senescence. A549 and HCT116 cells were treated for 7 days with concentrations of 50 nM doxorubicin, 2 nM disorazole C1, 25 nM discodermolide, or 2 nM vinblastine that had equivalent growth inhibition. Cells were lysed, protein was isolated, and the protein expression was determined by Western blotting (Fig. 6B). All compounds increased the levels of cyclin dependent kinase inhibitor p21 and cyclin D. Lower levels of Ser780 phosphorylated pRb and total pRb levels were also observed as expected for cells undergoing cell cycle arrest and premature senescence. Vinblastine (2 nM) also induced p53, p21, Rb, and cyclin D consistent with cell cycle arrest and senescence, although little β-galactosidase staining was observed (Figs. 6 and 7).

Fig. 7.

Disorazole C₁ causes premature senescence. A549 cells were treated for 7 days with vehicle (DMSO), doxorubicin (DOX; 50 nM), the inactive analog 010A (10 nM), discodermolide (DISCO; 25 nM) vinblastine (VBL; 2 nM), or disorazole C₁ (C₁, 1-10 nM) followed by β-galactosidase staining.

Further documenting the role of microtubule disruption in premature senescence, we observed DNA synthesis as measured by BrdU incorporation was inhibited by 67% in A549 cells treated for 1 day with IC₅₀ concentration of disorazole C₁ (2 nM), which was equivalent to that seen with the 50% growth-inhibitory concentration of doxorubicin (50 nM) (Supplemental Fig. 1; Supplemental Table 1). In contrast, almost all cells incorporated BrdU after treatment with the inactive analog O1OA, vinblastine, discodermolide, or DMSO vehicle control, indicating actively dividing cells. By 7 days, almost no cells were actively dividing after treatment with doxorubicin, discodermolide, or disorazole C₁ (≥0.5 nM) (Supplemental Fig. 1; Supplemental Table 1). Some decrease in BrdU-positive cells was seen after a week of treatment with either 10 nM O1OA or 2 nM vinblastine. We also excluded the possibility that the induction of premature senescence was the product of double-strand DNA breaks because we did not detect an increase in Ser139 phosphorylation of γH2AX after treatment with either disorazole C₁ or vinblastine (Supplemental Fig. 2). These results, in combination with the positive β-galactosidase staining and changes in cell cycle protein markers, indicate that disorazole C₁ could induce premature senescence in some human tumor cells.

Discussion

The highly electrophilic divinyl oxirane moiety of disorazole A₁ has been an attractive candidate mediator for the inhibition of tubulin polymerization, and our results indicate that at least some of the pharmacological activity of the disorazole family resides in other aspect of the disorazole pharmacophore. The disorazole C₁ concentration required to inhibit tubulin polymerization by 50% in vitro (namely, 11.8 μM) was approximately an order of magnitude higher than the IC₅₀ values reported for disorazole A₁ (1.8 μM) using a similar assay (Elnakady et al., 2004). Our in vitro tubulin binding studies indicated that disorazole C₁ had a unique direct tubulin binding interaction, possibly binding at or near a site typically occupied by vinblastine and dolastatin 10. However, we could not rule out the possibility that disorazole C₁ bound to a completely unrelated site on tubulin, instigating a conformational change that allosterically interferes with vinblastine and dolastatin 10 binding. Further studies are needed to determine the exact binding domain of disorazole C₁ on tubulin. These results indicate disorazole C₁, like disorazole A₁ (Elnakady et al., 2004), inhibited tubulin polymerization in the absence of abundant microtubule-associated proteins.

Disorazole C₁ prevented proper assembly of microtubules within cells. The disorazole C₁ concentration required to inhibit tubulin polymerization by 50% in vitro would be expected to be higher because the concentration of purified tubulin (10 μM) used in the in vitro reactions was intentionally high to permit robust detection. Moreover, it is now recognized that most effective anticancer drugs affect microtubule depolymerization or polymerization at much lower concentrations than required for in vitro studies (Jordan and Wilson, 2004). The lower concentration of disorazole C₁ or A₁ required to disrupt cellular proliferation probably reflected the need to affect only a small fraction of the tubulin/microtubule dynamics to block the cell division machinery. The breakdown of microtubules was seen as early as 15 min after a 5 nM disorazole C₁ treatment, with complete absence of cells with visible microtubules within 4 h (data not shown). We speculate that the long-lived disorazole C₁ cytotoxicity might be because of misalignment of chromosomes at the metaphase plate, consistent with the G₂/M arrest and increase in mitotic index that we described previously (Wipf et al., 2006). All of these data indicated that disorazole C₁ was a potent, cytotoxic, microtubule destabilizer that prevents normal cell division. Nonetheless, the previously reported IC₅₀ values for growth inhibition with disorazole A₁ suggest that it is even more potent than disorazole C₁, probably because of the highly reactive divinyl oxirane moiety. In preliminary growth inhibition studies with HeLa cells, we observed an IC₅₀ of 47 pM for disorazole A₁ compared with an IC₅₀ of 219 pM for disorazole C₁. It remains to be determined whether disorazole C₁ has a better therapeutic potential compared with the more potent disorazole A₁.

Proliferating cells that are subjected to significant harmful stress rapidly exit the cell cycle through several concurrent failsafe “checkpoint” mechanisms enabling DNA repair and suppressing possible tumorigenesis. If these cell cycle check-points are deficient because of the presence of mutations or overwhelming damage, cells generally engage in apoptosis or senescence. Senescence is defined as permanent cell-cycle arrest preventing proliferation but permitting metabolic activity. Senescent cells have an enlarged, flattened morphology with an increased cytoplasmic area, increased granularity, extensive cytoplasmic vacuoles, and multinucleation (Schmitt, 2007). There currently is no single accepted biochemical defining marker for the senescence phenotype, although perinuclear lysosomal presence of β-galactosidase is often considered to be a senescence characteristic (Dimri et al., 1995). In addition, senescent cells do not synthesize DNA and therefore display decreased BrdU uptake. Cells undergoing senescence show changes in the expression of genes involved in tumor suppression, cell cycle progression, mitosis, and DNA replication. Thus, the tumor suppressor protein p53, which up-regulates the cyclin-dependent kinase inhibitors p21 and p16, is activated and promotes growth arrest in cells undergoing senescence (Stein et al., 1999; Roninson, 2003; Serrano, 2007). The proliferation blockage associated with senescence seems to be irreversible because few cells actually recover from prolonged up-regulation of tumor suppressors, such as p53, p21, and p16.

Chemotherapy-induced senescence has been reported with mechanistically distinct agents including DNA alkylators, antimetabolites, topoisomerase inhibitors, irradiation, and some microtubule stabilizers (Chang et al., 1999; Klein et al., 2005; Schmitt, 2007). Senescence is typically induced at low drug concentrations, whereas high concentrations produce overwhelming damage that favors apoptosis. For example, treatment of human hepatoma cells with a low concentration of the indirect DNA-damaging agent doxorubicin (<100 nM) induces senescence, eventually leading to mitotic catastrophe, whereas a high concentration (20 μM) leads to apoptosis (Eom et al., 2005). Considering the broad mechanisms of action of the anticancer agents causally involved in tumor cell senescence, it is surprising that in initial studies, clinically important inhibitors of tubulin polymerization, such as vinblastine and vincristine, were not observed to be potent inducers of senescence (Chang et al., 1999; Schmitt, 2007), even though one would hypothesize that kinetochore nonattachment recognized by mitotic spindle checkpoints might lead to senescence (Yamada and Gorbsky, 2006). It is interesting that recent work by Arthur et al. (2007) suggests that the inhibitor of tubulin polymerization, JG-03-14 (Arthur et al., 2007), caused senescence in residual surviving cell populations. Treatment of L929 mouse fibroblasts and PtK2 rat kangaroo kidney epithelial cells for 5 days with disorazole A₁ caused enlarged cells with increased metabolic activity that ceased to divide, possibly indicating that these surviving cells adopted a senescence phenotype (Elnakady et al., 2004). We found positive β-galactosidase staining in cells treated with disorazole C₁ at concentrations well below the IC₅₀. Prominent β-galactosidase staining was not observed previously with vinblastine (Lee et al., 2006), which no doubt contributed to the current belief that inhibition of microtubule polymerization does not provide a signal for senescence (Schmitt, 2007). We also did not observe significant β-galactosidase expression in cells treated with vinblastine or vincristine at their IC₅₀ concentrations for growth inhibition. It is also important to note, however, that β-galactosidase expression has been questioned as a marker for senescence (Lee et al., 2006). Thus, we examined additional markers, including increased p53 and p21 with down-regulation in total Rb and phospho-Rb consistent with senescence. A549 cells do not express another common marker of senescence, p16^INK4; rather, they use p21 to maintain cell cycle arrest and senescence (Kashiwabara et al., 1998; McConnell et al., 1998). We saw increases in the levels of cyclin D, and the cyclin-dependent kinase inhibitor p21 was also elevated. It has been suggested that this phenomenon occurs because of hypermitogenic senescence, where supraphysiologic levels of mitogens (Ras/Raf/mitogen-activated protein kinase) increase cyclin D levels (Blagosklonny, 2003), but to date, our efforts to identify activation of the mitogen-activated protein kinase pathway have not been successful. Finally, another marker of senescence is inhibition of DNA synthesis, which was markedly decreased in cells exposed to disorazole C₁ as early as 24 h after treatment with low concentrations (≥0.5 nM); both discodermolide and vinblastine were less efficacious.

In summary, our data provide evidence that microtubule destabilization can suppress cancer cell growth by inducing premature senescence at concentrations at or below the IC₅₀ for growth inhibition. Other microtubule-destabilizing agents, namely JG-03-14, also are reported to cause premature senescence (Arthur et al., 2007), albeit at higher concentrations (namely 500 nM, which causes 70-80% growth inhibition) than that required with disorazole C₁. These results illustrate that microtubule disruption, like some other forms of cellular stress, provided intracellular signals promoting premature senescence, at least in some tumor cell types. The propensity of disorazole C₁ to promote senescence may be related to the prolonged cellular effects or the nature of its interactions with microtubules. The symmetric synthesis of disorazole C₁ (Wipf and Graham, 2004; Wipf et al., 2006) enables the production of radiolabeled compound and analog development. We speculate that disorazole C₁ could be a potent and valuable chemical probe for studying the process of premature senescence induced by mitotic spindle disruption.