Background
Griseofulvin (GF), an orally active antifungal drug, has been attracting considerable interest as a potential anticancer agent owing to its low toxicity and efficiency in inhibiting the proliferation of different types of cancer cells [
1‐
4]. GF in combination with nocodazole was shown to potently inhibit tumor growth in athymic mice [
1]. It induces apoptosis in several cancer cell lines [
5] and it has also been proposed that GF can selectively kill the cancer cells sparing the normal healthy cells [
3].
GF is known to inhibit the growth of fungal, plant and mammalian cells mainly by inducing abnormal mitosis and blocking the cells at G2/M phase of cell cycle [
1‐
3,
6‐
8]. Different organisms exhibit different degrees of sensitivity to GF owing to its differential affinity to different tubulins [
6,
9]. The concentration required to inhibit the growth of fungal cells is much lower than that required to inhibit the mammalian cells due to its higher affinity for fungal tubulin as compared to the mammalian tubulin [
6,
10‐
12]. GF has been reported to interact with tubulin [
2,
12‐
16] as well as microtubule associated proteins (MAPs) [
13,
17].
Recently, GF has been shown to suppress the dynamic instability of MAPs-free microtubules
in vitro [
2]. The spindle microtubules of HeLa cells treated with moderate concentrations of GF appeared to have nearly normal organization [
2,
18], while higher GF concentration caused depolymerization of the microtubules [
2,
16]. Based on the strong suppressive effects of GF on the microtubule dynamics
in vitro, it was proposed that GF inhibits mitosis in HeLa cells by suppressing microtubule dynamics [
2].
Although several studies suggested that tubulin is the primary target of GF [
2,
12,
14‐
16], the binding site of GF in tubulin is yet unknown. Based on the findings that GF quenches tryptophan fluorescence of the colchicine-tubulin complex [
12] and that colchicine can depolymerize GF-induced polymers of tubulin in the presence of MAPs at 4°C [
13], it was suggested that GF binds at a site distinct than the colchicine binding site in tubulin [
12].
In this study, we have identified two potential binding sites for GF in mammalian tubulin and provided a mechanistic explanation of how GF stabilizes microtubule dynamics. Further, the data suggested that GF inhibited mitosis in MCF-7 cells by suppressing the dynamicity of microtubules and that a population of the mitotically blocked cells escaped mitosis with misegregated chromosomes and eventually underwent apoptotic cell death.
Methods
Materials
GF, paclitaxel, vinblastine, mouse monoclonal anti-α tubulin IgG, rabbit monoclonal anti-γ tubulin IgG, alkaline phosphatase conjugated anti-mouse IgG, FITC (fluorescein isothiocyanate) conjugated anti-mouse IgG, fetal bovine serum, bovine serum albumin and Hoechst 33258 were purchased from Sigma (St. Louis, MO, USA). Mouse monoclonal anti-BubR1 antibody was purchased from BD Pharmingen (San Diego, USA). Rabbit polyclonal anti-Mad2 IgG was purchased from Bethyl laboratories (Montgomery, USA). Mouse monoclonal anti-Hec 1 IgG was purchased from Abcam (Cambridge, MA, USA). Anti-Mouse IgG alexa 568 conjugate and lipofectamine-2000 were purchased from Invitrogen (Carlsbad, CA, USA). Mouse monoclonal anti-p53 IgG, mouse monoclonal anti-p21 IgG, rabbit polyclonal anti-phosphohistone IgG and Annexin V apoptosis detection kit were purchased from Santa Cruz Biotechnology (CA, USA). All other reagents were of analytical grade.
Effects of GF on MCF-7 cells and on the cell cycle progression
MCF-7 cells (1 × 10
5 cells/mL) were grown in 96-well tissue culture plates at 37°C for 24 h [
19]. Then the medium was replaced with fresh medium containing vehicle (0.1% DMSO) or different concentrations of GF and the cells were grown for additional 48 h. Both attached and floating cells were harvested with the help of trypsin-EDTA solution and counted after staining with trypan blue [
2]. MCF-7 cells were grown on glass coverslips in the absence or presence of different concentrations of GF for 24 h and the mitotic index was calculated by staining the chromosomes with Hoechst dye [
19,
20].
For determining the effect of GF on cell cycle progression, MCF-7 cells were grown for 48 h (one cell cycle) without or with different concentrations of GF. Samples were prepared as described recently [
21]. DNA content of the cells was quantified in a flow cytometer (FACS Aria special order system, Becton Dickinson) and the cell cycle distribution was analyzed using the Modfit LT program [
21].
Immunofluorescence Microscopy
MCF-7 cells (5 × 10
4 cells/mL) were grown on glass coverslips in 24 well tissue culture plates for 24 h [
19,
20]. Then, the medium was replaced with fresh medium containing vehicle (0.1% DMSO) or different concentrations of GF (15, 30, 60 and 90 μM) and the incubation continued for further 24 or 48 h. The cells were then fixed with 3.7% formaldehyde at 37°C for 30 min and processed to visualize α tubulin, γ tubulin, p53, p21, BubR1, Mad2, Hec1 and phosphohistone [
19,
20].
Analysis of the polymeric mass of tubulin
MCF-7 cells were incubated without or with different concentrations of GF (15, 30, 60 and 90 μM) for 48 h. The effect of GF on the amount of polymerized tubulin in MCF-7 cells was estimated by western blot analysis using anti α-tubulin antibody as described earlier [
19] and the band intensities were estimated using Image J software. The band intensities for polymeric and soluble tubulin in the presence of different concentrations of GF were normalized with that of the vehicle-treated MCF-7 cells. The normalized band intensities were plotted against GF concentrations.
Transfection of EGFP-α tubulin and measurement of microtubule dynamics
The effects of GF on the dynamic instability of individual microtubules in live MCF-7 cells were determined using EGFP-α tubulin as described recently [
19,
21]. Briefly, MCF-7 cells stably-expressing EGFP-α tubulin were incubated with vehicle or different concentrations of GF (5 and 15 μM) for 24 h. Individual microtubules on the peripheral region of the MCF-7 cells were observed using a 60 × water immersion objective in a FV 500 laser scanning confocal microscope (Olympus, Tokyo, Japan) [
19,
21]. The images were digitally magnified by 6 times and were acquired at 2 or 4 s intervals for a total period of 120-180 s using the FluoView software (Olympus, Tokyo, Japan). The lengths of the microtubules at different time points were calculated using Image-Pro Plus software (Media Cybernetics, Silver Spring, MD). The dynamic instability parameters were calculated as described earlier [
2,
19]. Transfection of MCF-7 cells with p53-GFP plasmid was done using lipofectamine-2000. Stably expressing p53-GFP MCF-7 cells were treated with different concentrations of GF (15-90 μM) and the localization of p53-GFP was studied using Nikon Eclipse TE-2000U fluorescent microscope. The images were analyzed with Image-Pro Plus software.
Annexin V/propidium iodide staining
MCF-7 cells were grown in the absence and presence of different concentrations of GF (15, 30, and 60 μM) for 48 h. Control and treated cells were stained using the Annexin V/propidium iodide apoptosis kit, as described earlier [
19] and the cells were examined under a fluorescence microscope.
Effect of GF on BODIPY FL-vinblastine binding to tubulin
Goat brain tubulin (2 μM) was incubated without and with 100 μM GF, and with 50 μM vinblastine at 37°C for 45 min. Then, 2 μM fluorescent-tagged vinblastine (BODIPY FL-vinblastine) was added to each of the reaction mixtures and incubated for an additional 20 min in dark at 25°C. Fluorescence spectra of BODIPY FL-vinblastine were monitored by exciting at 490 nm.
Effect of paclitaxel on GF binding to tubulin
Tubulin (10 μM) was polymerized in presence of 1 mM GTP, 5 mM MgCl
2, and 10% DMSO without and with 15 μM paclitaxel at 37°C for 10 min. GF (50 and 100 μM) was added to the polymerization mixtures and then tubulin was allowed to polymerize for an additional 20 min. The polymers were pelleted down and washed with warm pipes buffer. The polymers were depolymerized on ice for 1 h using 0.01% NaOH in 25 mM pipes buffer. Fluorescence of GF was monitored at 437 nm by excitation at 330 nm. The concentration of GF in the pellet was then calculated using the fluorescence values of known concentrations. The concentration of protein in the pellet was measured using Bradford method [
22].
Docking of GF to tubulin
GF structure was obtained from the drugbank [
23]. Protein coordinates were obtained from the protein data bank [
24]. Modeler9v4 [
25] was used for 3D structure modeling. The LigandFit [
26] module of Discovery Studio 2.1 (Accelrys Inc., California) and Autodock4 [
27] were used for docking. Naccess V2.1.1 was used for calculating solvent accessible surface area [
28]. PyMol was used for rendering 3D structures [
29].
The crystal structure of tubulin complexed with epothilone A [
30] (pdb id 1TVK) was used for docking since this has the highest resolution (2.89 Å) among all the tubulin structures available to date. However, in this structure, residues 35A-60A (the suffix denotes the subunit) are disordered. This region was modeled using the corresponding region of β tubulin as the template. The (Φ,Ψ) values of all the residues in the modeled region are within the allowed region of the Ramachandran map [
31]. In addition, the WHAT IF server [
32] was used to check the modeled structure (Additional file
1, Table S1). Docking was performed for epothilone A as a control, using both the softwares. The root mean square deviation (RMSD) between the "predicted" and experimentally determined binding modes of epothilone A is 2 Å.
Docking using Autodock4
Blind docking was performed by treating only GF as flexible. The entire protein was enclosed in a box with a grid spacing of 0.375 Å. Fifty docking jobs, each of hundred runs, were carried out using the Lamarckian genetic algorithm. Default values were used for all the parameters; g_eval was set to 2,500,000 (M). The resulting 5000 binding modes were clustered using an all-atom RMSD cutoff of 2 Å. Binding modes are ranked on the basis of cluster size. Highly populated (viz., size ≥ 50) clusters were chosen for further analysis. In these clusters, decrease in solvent accessible surface area of the protein on GF binding was calculated.
Docking using LigandFit
Binding sites were defined on tubulin surface and GF was docked in each of these cavities. The ligand poses at these sites were selected on the basis of shape matching, using a cutoff distance of 3 Å between the ligand and the binding site residues. Binding modes are ranked on the basis of consensus scoring.
Determination of combination Index (CI)
MCF-7 cells were treated with GF (10 and 15 μM), vinblastine (0.5 and 1 nM) or GF along with vinblastine for 48 h. The CI was calculated by Chou and Talalay method [
33‐
35] using the equation:
Where, (D)1 and (D)2 are the concentrations of drug 1 (GF) and drug 2 (vinblastine) in combination that produces a given effect, (Dx)1 and (Dx)2 are the concentrations of drug 1 and drug 2 that also produces the same effect when used alone.
The concentration of the drug that produces a particular effect
(Dx) and the median dose (
Dm) were calculated as described earlier [
33‐
35]. CI <1 indicates synergism, CI = 1 indicates additivity, and CI >1 indicates antagonism.
Discussion
GF (≤ IC
50) strongly suppressed the dynamics of individual microtubules in live MCF-7 cells without detectably altering the microtubule network. However, at higher concentrations, GF induced significant depolymerization of both the mitotic spindles and the interphase microtubules. The suppressive effects of GF on the dynamic instability of interphase microtubules of MCF-7 cells were found to be qualitatively similar to its effects on the bovine brain microtubules
in vitro [
2]. In our studies, the IC
50 of GF for the inhibition of cell proliferation has been found to be 17 ± 2 μM, which is comparable to some of the anticancer agents that are undergoing clinical trials. For example, estramustine (clinical trials.gov identifier NCT00151086), curcumin (NCT00094445) and noscapine (NCT00912899) inhibit the proliferation of MCF-7 cells with the IC
50 of 5 ± 1 μM [
21], 12 ± 0.6 μM [
39] and 39.6 ± 2.2 μM [
40], respectively.
GF was also found to be incorporated into the microtubules in high stoichiometry (0.24 molecules of GF per tubulin dimer) suggesting that GF binds along the length of the microtubules. Like paclitaxel [
41,
42], GF did not strongly influence the time based catastrophe and rescue frequencies. GF suppressed the dynamics by reducing the rate and extent of the growing and shortening excursions and increasing the time microtubule spent in the pause state. The docking studies indicated that GF has two potential binding sites in tubulin; one of these sites is overlapping with the paclitaxel binding site and the other lies at the interface of αβ tubulin, which is distinct from the GDP, vinblastine, and colchicine sites. The docking analysis is consistent with the findings that GF neither binds to the colchicine site [
12] nor the vinblastine site in tubulin (Additional file
1, Figure S5; Additional file
1, Table S5). A competition experiment with paclitaxel showed that paclitaxel reduced the binding of GF to tubulin in microtubules supporting the computational analysis data that GF binding site partially overlaps with the paclitaxel site in tubulin. Paclitaxel is known to stabilize microtubule dynamics [
42] and like GF, paclitaxel has been shown to bind along the length of microtubules [
41]. Therefore, it is logical to propose that GF stabilizes microtubules dynamics by binding to tubulin in the paclitaxel site. It is likely that at higher concentrations GF binds to tubulin in the putative second site, which is located at the intra-dimer interface and induces microtubule depolymerization.
Defects in the microtubule-kinetochore attachment and the tension across the sister kinetochores are sensed by the check point proteins and the accumulation of check point proteins at the kinetochore region is thought to prevent the cells to enter into the anaphase until the defects are corrected [
20,
21,
36,
37]. The enhanced localization of BubR1 and Mad2 on the kinetochores upon GF treatment suggested that GF inhibited the extinction of checkpoint proteins from the kinetochores (Figure
1C; Additional file
1, Figure S2). The accumulation of BubR1 and Mad2 at the kinetochores activated the mitotic checkpoint and arrested the cells at mitosis. These mitotically blocked cells either undergo apoptosis or make an aberrant mitotic exit without cytokinesis resulting in cells with fragmented nuclei, which eventually undergo apoptosis [
3,
43]. The presence of multiple poles in GF treated cells probably results in improper chromosome segregation leading to the formation of multiple nuclei. The FACS analysis did not show an increase in DNA content (> 4N) indicating that multiple nuclei in the cells are indeed due to the improper chromosome segregation and not due to the multiplication of DNA. GF treatment caused a strong increase in multipolar mitosis leading to the formation of fragmented nuclei of varying sizes in MCF-7 cells. Most (> 60%) of these cells with fragmented nuclei had much higher accumulation of p53 as compared to the mononucleated cells suggesting that the cells that committed aberrant exit from the mitotic block with fragmented nuclei underwent apoptosis. Therefore, GF may induce apoptosis in MCF-7 cells via a series of concerted events, which includes formation of multipolar spindles, fragmentation of the nucleus, nuclear accumulation of p53, and finally p53 dependent induction of apoptosis. Cells treated with higher GF concentrations (> 30 μM) had hyper amplified centrosomes (Figure
3B; Additional file
1, Figure S4) and completely disorganized multipolar mitosis (Figure
3B). As a result, GF induced a concentration and a time dependent increase in the number of cells containing fragmented nuclei. The organization of centrosomes plays an important role in the successful completion of mitosis. Microtubule interacting drugs like paclitaxel, nocodazole, vinblastine and podophyllotoxin were shown to affect the organization of the centrosomes and cause functional impairment [
44]. It has been found that GF inhibited the centrosomal clustering without interfering with the functions of NuMA and dynein and it was indicated that the alteration of the interphase microtubule stability by GF might be the reason for the inhibition of centrosomal clustering [
3]. The evidence presented in this study strongly suggested that the kinetic suppression of microtubule dynamics induces mitotic irregularities and nuclear accumulation of p53.
Drugs having adverse side effects can be successfully used for chemotherapy if their effective doses are reduced significantly. Moreover, the use of combination of two or more drugs reduces the chances of survival of the resistant cancer cells [
44]. For example, the use of haloperidol in combination with vinblastine reversed the resistance of K562/VBL cells to vinblastine [
45]. In this work, we have found that the combination of GF and vinblastine exhibited strong synergistic effects on the inhibition of proliferation of MCF-7 cells.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
KR performed microtubule dynamics studies and the cell culture experiments, analyzed the data and contributed in writing the manuscript. BJ performed docking studies and in vitro experiments and contributed in writing the manuscript. JA performed the cell culture experiments and contributed in writing the manuscript. PS performed flow cytometry and contributed in manuscript preparation. PVB provided help for docking studies, manuscript preparation and scientific discussions. DP provided the resources for the work, helped in data analysis and wrote the manuscript. All the authors have read and approved the final version of the manuscript.