Diallyl Trisulfide Suppresses the Proliferation and Induces Apoptosis of Human Colon Cancer Cells through Oxidative Modification of β-Tubulin*

Allyl sulfides are characteristic flavor components obtained from garlic. These sulfides are thought to be responsible for their epidemiologically proven anticancer effect on garlic eaters. This study was aimed at clarifying the molecular basis of this anticancer effect of garlic by using human colon cancer cell lines HCT-15 and DLD-1. The growth of the cells was significantly suppressed by diallyl trisulfide (DATS, HCT-15 IC50 = 11.5 μm, DLD-1 IC50 = 13.3 μm); however, neither diallyl monosulfide nor diallyl disulfide showed such an effect. The proportion of HCT-15 and that of DLD-1 cells residing at the G1 and S phases were decreased by DATS, and their populations at the G2/M phase were markedly increased for up to 12 h. The cells with a sub-G1 DNA content were increased thereafter. Caspase-3 activity was also dramatically increased by DATS. Fluorescence-activated cell sorter analysis performed on the cells arrested at the G1/S boundary revealed cell cycle-dependent induction of apoptosis through the transition of the G2/M phase to the G1 phase by DATS. DATS inhibited tubulin polymerization in an in vitro cell-free system. DATS disrupted microtubule network formation of the cells, and microtubule fragments could be seen at the interphase. Peptide mass mapping by liquid chromatography-tandem mass spectrometry analysis for DATS-treated tubulin demonstrated that there was a specific oxidative modification of cysteine residues Cys-12β and Cys-354β to form S-allylmercaptocysteine with a peptide mass increase of 72.1 Da. The potent antitumor activity of DATS was also demonstrated in nude mice bearing HCT-15 xenografts. This is the first paper describing intracellular target molecules directly modified by garlic components.

nents of the essential oil prepared from garlic (Allium sativum L.). Garlic is widely served around the world, and it has been reported that allyl sulfides inhibit both the initiation and promotion stages of tumorigenesis in experimental carcinogenesis models for various types of cancer (1)(2)(3)(4)(5). Recently, several lines of investigation have shown that allyl sulfides suppress cell growth and induce apoptosis in multiple cancer cell lines (6 -12). We previously reported that the sulfur-containing volatile oils prepared from garlic and onion inhibit proliferation and induce differentiation of the human promyelocytic leukemia cell line HL-60 (13). However, the molecular mechanisms underlying the antitumorigenesis of allyl sulfides are still not fully understood.
Microtubules are ubiquitous proteins present in eukaryotes as components of the cytoskeleton and play pivotal roles in a variety of cellular processes involving cell division, motility, and intracellular trafficking (14). The microtubules are dynamic polymers composed of ␣␤-tubulin heterodimers, and they form the mitotic spindles, which are known to introduce the replicated DNA molecules to the respective daughter cell. Thus, antimitotic drugs developed for targeting microtubules have gained great success in cancer chemotherapy (15,16). Various microtubule-interacting drugs, such as Vinca alkaloid and paclitaxel, cause mitotic arrest prior to the induction of apoptosis in tumor cells. It is quite natural that the suppression of spindle dynamics by these drugs hampers or completely blocks the mitosis of cells, especially at the transition from metaphase to anaphase.
This study was aimed at clarifying the molecular target of allyl sulfides to understand the anticancer mechanism elicited by the consumption of garlic. We show the structure-function relationship of allyl sulfides in the inhibition of human colon adenocarcinoma cell lines at first, and the changes in the cells caused by DATS, which include the disruption of spindle formation, sustainment of the cyclin B1 expression, mitotic arrest, and finally apoptosis. We also report for the first time that the direct modification of specific cysteine residues in ␤-tubulin molecules by DATS in vitro causes these events. Finally, we show the tumor growth inhibition by DATS in a xenograft mouse model in vivo.

MATERIALS AND METHODS
Cells and Chemicals-Human colon adenocarcinoma cell lines HCT-15 and DLD-1 (obtained from Cell Resource Center for Biomedical Research, Tohoku University, Sendai, Japan) were grown and maintained in RPMI 1640 medium (Invitrogen) supplemented with 10% fetal bovine serum (Sigma) at 37°C in 95% air and 5% CO 2 . DAS and DADS were purchased from Tokyo Kasei Kogyo Co., Ltd. (Tokyo, Japan). DATS was synthesized by using Bunte salt (17). DATS and commercial DADS were purified by high pressure liquid chromatography (Alliance 2695 system; Waters Co., Milford, MA) on an Inertsil ODS-3 column (6 mm ϫ 250 mm; GL Science, Tokyo, Japan).
Cell Proliferation Assay-HCT-15 and DLD-1 cells were precultured for 48 h and then exposed to DAS, DADS, or DATS for 24 h. The cells were then counted with a hemocytometer. The rate of growth inhibition by the allyl sulfides was calculated based on the control culture, which was treated with vehicle only (0.1% Me 2 SO), taken as 100% growth.
Cell Synchronization-HCT-15 and DLD-1 cells were synchronized at the G 1 /S boundary by the method known as the double thymidinehydroxyurea block. The cells were presynchronized at the S phase by incubation with 2.5 mM thymidine for 12 h. Then they were released by changing the medium to the thymidine-free fresh medium and incubating them for 12 h, after which they were resynchronized at the G 1 /S transition point by incubation with 1 mM hydroxyurea for 12 h. The cells were washed and incubated in the fresh medium to re-enter the cell cycle.
Cell Cycle Analysis-The cell cycle distribution of HCT-15 and DLD-1 cells was measured by flow cytometry. The harvested cells (ϳ10 6 cells) were fixed with ice-cold 70% ethanol, treated with 500 g/ml RNase A (Sigma), and subsequently stained with 25 g/ml propidium iodide (Sigma). Then they were analyzed by using a flow cytometer FACSCalibur (Becton Dickinson, Franklin Lakes, NJ) and FlowJo software (Treestar Inc., Ashland, OR).
Western Blot Analysis of Cyclin B1-Total cellular protein (30 g) was subjected to sodium dodecylsulfate-10% polyacrylamide gel electrophoresis. The proteins separated were electrically transferred to a cellulose nitrate membrane (Advantec Toyo, Tokyo, Japan) for immunoblot analysis. The blot was incubated with mouse anti-cyclin B1 monoclonal antibody (1:2000; Upstate Biotechnology, Inc., Lake Placid, NY) for 1 h and then incubated with horseradish peroxidase-labeled anti-mouse secondary antibody (1:2000; Jackson ImmunoResearch Laboratories, West Grove, PA) for 1 h. The membrane was developed by using a commercial kit (HRP-1000 Immunostain; Konica, Tokyo, Japan).
Fluorometric Assay of Caspase-3 Activity-Cell lysates containing 30 g of protein were incubated for 60 min at 37°C in reaction buffer (20 mM PIPES, 100 mM NaCl, 1 mM EDTA, 0.1% CHAPS, 10 mM dithiothreitol, and 10% sucrose, pH 7.2) containing 50 M acetyl-L-aspartic-Lglutamic-L-valyl-L-aspartic acid ␣-(4-methylcoumaryl-7-amide) (Peptide Institute, Inc., Osaka, Japan). The 7-amino-4-methylcoumarin released was measured by use of a spectrofluorometer with excitation at 380 nm and emission at 460 nm. Tubulin Polymerization Assay-Tubulin was purified from pig brain by use of a phosphocellulose column and dissolved in PIPES buffer (1.5 mg tubulin/ml of 80 mM PIPES, pH 6.8, 1 mM MgCl 2 , 1 mM EGTA, 10% glycerol, and 1 mM GTP). The tubulin solution was then placed in a thermostatically controlled cuvette at 4°C for 10 min in the presence or absence of DATS. To initiate tubulin polymerization, the reaction mixture was warmed at 37°C. The tubulin polymerization was monitored by measuring the increase in the absorbance at 340 nm.
Indirect Immunofluorescence Microscopy-The cells were cultured on a Thermanox coverslip (Nalge Nunc International, Rochester, NY) and fixed with acetone/methanol (1:1) for 2 min at room temperature. After washing with phosphate-buffered saline, the fixed cells were incubated with mouse anti-␤-tubulin monoclonal antibody (1:500; Sigma) for 30 min at room temperature, followed by incubation with Alexa Fluor 488 goat anti-mouse IgG antibody (1:500; Molecular Probes, Inc., Eugene, OR) for 30 min. The specific fluorescence was observed by a confocal microscope (Fluoview; Olympus, Tokyo, Japan).
High Performance Liquid Chromatography-Tandem Mass Spectrometry-Phosphocellulose-purified tubulin (1 mg/ml) was incubated at 37°C for 60 min in the presence or absence of 100 M DATS. The DATS-treated and native tubulin were digested with modified trypsin (Promega, Madison, WI) and analyzed by liquid chromatography-tandem mass spectrometry by using a MAGIC C18 column (0.15 mm ϫ 50 mm; Michrom Bioresources, Auburn, CA). The peptides were eluted over a 20-min period with a linear gradient 5-65% in terms of solvent B going from solvent A (2% (v/v) acetonitrile, 0.1% formic acid) to solvent B (90% (v/v) acetonitrile, 0.1% formic acid) with a flow rate of 0.8 l/min. The tryptic peptide samples were separated and analyzed with a LCQ Deca XP ion trap mass spectrometer (ThermoFinnigan, San Jose, CA). Tandem mass spectrometry data obtained were analyzed by using SEQUEST, a computer program that allows the correlation of experimental data with theoretical spectra generated from known protein sequences (18). In this work, the general criteria for a preliminary positive peptide identification for a doubly charged peptide were a correlation factor greater than 2.5, a delta cross-correlation factor greater than 0.1 (indicating a significant difference between the best match reported and the next best match), a high preliminary scoring, and a minimum of one tryptic peptide terminus. For triply charged peptides, the correlation factor threshold was set at 3.5. All of the matched peptides were confirmed by visual examination of the spectra. All of the spectra were searched against the data in FASTA format generated from pig ␣-tubulin (NCBI accession number P02550) and pig ␤-tubulin (NCBI accession number P02554) in the data base of National Center for Biotechnology Information.
Measurement of Cysteine Residues in Tubulin-The number of cysteine residues in the tubulin was determined by titrating the sulfhydryl group in tubulin with 5,5Ј-dithiobis-2-nitrobenzoic acid. Phosphocellulose-purified tubulin (0.1 mg/ml) was incubated with 10 M DATS or vehicle at 25°C for 20 min. After the incubation, tubulin samples were mixed with 5,5Ј-dithiobis-2-nitrobenzoic acid (1 mM), and the absorbance at 412 nm was measured. The number of cysteine residues was calculated from the standard curve drawn by using cysteine (Wako Pure Chemical, Osaka, Japan).
Antitumor Effect of DATS on Mice Xenograft Model-All of the animal experiments were performed in accordance with the Guidelines for Animal Experiments of the College of Bioresource Sciences at Nihon University. Tumor xenografts were maintained by serial subcutaneous transplantation of 2 ϫ 2 ϫ 2-mm fragments of HCT-15 tumor into the right subaxillary region of 6-week-old female athymic CAnN.Cg-Foxn1 nu /CrlCrlj mice (Charles River Laboratories Japan, Inc., Yokohama, Japan) on day 0. DATS (dissolved in 90% saline, 5% ethanol, and 5% Cremophore EL; Sigma, 6 mg/kg) or vehicle was injected into a tail vein on days 7, 10, 13, 16, 19, 22, and 25 (every 3 days, 7 doses in total). Body weight and tumor sizes were measured every 5 days. Tumor volume was calculated by the equation V ϭ [L ϫ W 2 ] ϫ 0.52 (where V is volume, L is length, and W is width). The tumors excised were fixed in 4% paraformaldehyde/phosphate-buffered saline for 48 h at 4°C, embedded in paraffin, sectioned in 1.5-m-thick sections, and stained with hematoxyline and eosine.

Inhibition of the Growth of Human Colon Cancer Cells by Allyl Sulfides in Vitro-
We initially examined the effect of allyl sulfides (structures shown in Fig. 1) on the growth of HCT-15 and DLD-1 cells. The cell growth was significantly reduced by DATS in a concentration-dependent manner (IC 50 ϭ 11.5 Ϯ 0.8 M for HCT-15, IC 50 ϭ 13.3 Ϯ 0.7 M for DLD-1). In contrast, neither DAS nor DADS showed any effect on both cell lines. Microscopic observation revealed that most HCT-15 cells in the control cultures were in interphase, with only a small fraction (1-2%) of them in the M phase (Fig. 2B, 0 h). On the contrary, HCT-15 cells cultured in the presence of DATS exhibited mitotic arrest as early as 6 h. More than 60% of the cells were arrested at the M phase by the treatment with DATS for 12 h, and the mitotic morphology was observed for up to 18 h (Fig. 2B, 6, 12, and 18 h). In these cells, characteristic changes such as chromosome aggregation and disappearance of nuclear membrane were observed.
Inhibition of Cell Cycle Progression and Induction of Apoptosis by DATS-The effect of DATS on the cell cycle progression of HCT-15 and DLD-1 cells was analyzed further by using flow cytometry. Under a normal condition in the absence of DATS, the cell cycle distribution of most HCT-15 and DLD-1 growing asynchronously was as follows: HCT-15 (51% in G 1 phase, 35% in S phase, and 14% in G 2 /M phase) and DLD-1 (69% in G 1 phase, 31% in S phase, and 9% in G 2 /M phase; Fig. 3A, 0 h). When DATS was added to the cultures, the percentages of HCT-15 and DLD-1 cells at G 1 and S phase decreased, and the cell population at G 2 /M phase markedly increased in a time-dependent manner. HCT-15 cells at 12 h after the addition of DATS, ϳ70% of the cells were at the G 2 /M phase (Fig. 3A, 12 h). The population of DLD-1 cells in the G 2 /M phase at 16 h after the addition of DATS reached maximum (Fig. 3A, 16  h). The cells with a sub-G 1 DNA content, which is an indicator of apoptosis, appeared at 12-16 h and increased thereafter in HCT-15 and DLD-1 cultures treated with DATS ( Fig. 3A, 12, 16, 20, and 24 h). Cyclin B1, a protein known to increase during the transition from the G 2 through the M phase, was accumulated at 6 -12 h after the DATS treatment (Fig. 3B). The time of expressing cyclin B1 in DATS-treated DLD-1 cells was slightly later than that in DATS-treated HCT-15 cells. This delay might be due to the difference in the doubling time of these cells. Caspase-3 activity also dramatically increase after the cells arrested at the G 2 /M phase (Fig. 3C). Hoechst 33258 staining of HCT-15 cells demonstrated that DATS treatment caused more chromatin condensation and nuclear fragmentation than found in the vehicle-treated control cells (Fig. 3D).
Induction by DATS of Cell Cycle-dependent Apoptosis through the Transition of G 2 /M to G 1 Phase-To determine whether the DATSinduced apoptosis originated from a specific stage of the cell cycle, we synchronized HCT-15 and DLD-1 cells at the G 1 /S boundary by the thymidine-hydroxyurea double-block method. After arrest of the cells at the G 1 /S boundary, the culture medium was replaced with fresh medium containing vehicle (thymidine/hydroxyurea 3 vehicle) or DATS (thymidine/hydroxyurea 3 DATS). In HCT-15 cells, after releasing from the block at G 1 /S boundary, the vehicle-treated cells went into the late S phase at 4 h and the G 2 /M phase at 6 h (Fig. 4A). At 8 h post-release, the vehicle-treated cells already at the G 2 /M phase entered into the next G 1 phase. Most of the cells were located in the G 1 phase of the next cycle by 10 h after replacement of the culture medium. On the contrary, the cells released from the G 1 /S block in the presence of DATS exhibited a delayed cell cycle progression. At 8 h, 80% of the DATS-treated cells were in the G 2 /M phase. At 10 -12 h, 60% of the cells still remained in the G 2 /M phase. After 10 h, the cells at sub-G 1 markedly increased in number in a time-dependent manner. DLD-1 cells showed more delayed cell cycle progression than HCT-15 cells (Fig. 4B). DATS did not induce apoptosis of HCT-15 cells arrested at the G 1 /S phase, and the cells with a sub-G 1 DNA content was 6.3% in the vehicletreated cells and 7.8% in the DATS-treated cells, respectively. There was no apparent morphological difference between these two groups. These results suggest that induction of apoptosis by DATS requires the cell cycle progression from the G 2 /M to the G 1 phase.
Inhibition of Tubulin Polymerization by DATS in a Cell-free System-Based on the finding that DATS inhibited the mitosis of HCT-15 cells (Fig. 2B), we next examined the effect of DATS on the polymerizationdepolymerization cycle of tubulin. Polymerization of phosphocellulosepurified tubulin was measured in the presence of glycerol and GTP as an increase in turbidity (absorbance at 340 nm). As shown in Fig. 5A, colcemid, a microtubule-depolymerizing agent, inhibited tubulin polymerization, whereas paclitaxel, a microtubule-stabilizing agent, enhanced it. DATS (10 M) also completely inhibited microtubule formation. Neither DAS (100 M) nor DADS (100 M) showed any effect on the microtubule formation (Fig. 5B). These data indicate that like a microtubuledepolymerizing agent, DATS inhibits tubulin polymerization in a cellfree system.

Disruption of Cytoplasmic Microtubule Organization by DATS-
The effect of DATS on the microtubule organization was examined by immunostaining of ␤-tubulin in DLD-1 human colon cancer cells. The normal microtubule distribution and its network formation were observed in the cytoplasm of vehicle-treated DLD-1 cells at interphase (Fig. 5C, Vehicle). Colcemid caused the disruption of microtubule net-  work formation, whereas paclitaxel did not show any apparent influence on microtubule network formation at interphase (Fig. 5C, Colcemid (3  h) and Paclitaxel (3 h)). DATS caused the disruption of microtubule network formation by depolymerization of the microtubules, and most cells had shorter microtubule fragments than those observed in the cells at interphase. The shorter microtubules were scattered throughout the cytoplasm of the cells at 3 h after the DATS treatment (Fig. 5C, DATS (3  h)). Because the mitotic spindle is a highly dynamic structure, it is susceptible to antimitotic agents. DATS inhibited spindle formation, and the nuclear membrane disappeared from the cells treated with DATS. Treatment of DLD-1cells with DATS caused the accumulation of the cells at prometaphase; the nuclear membranes of the cells disappeared, and there was no spindle formation (Fig. 5C, DATS (12 h)). When the test compounds were added after having reached the maximum turbid-ity, both DATS and colcemid caused the decrease in the turbidity. On the contrary, paclitaxel further increased the turbidity in the assay system. Taken together, DATS altered the microtubule structure by acting as a microtubule-depolymerizing agent.
Identification of DATS Modification Site in the Tubulin Molecule-Tubulin contains reactive sulfhydryl groups in its molecular structure (19). To examine whether DATS directly reacted with the tubulin molecule, we digested both DATS-treated and native tubulin samples with trypsin and analyzed the digests by liquid chromatography-tandem mass spectrometry. Peptide mass mapping of the DATS-treated tubulin identified 28.8% of it as ␣-tubulin peptides and 40.0% as ␤-tubulin peptides. The DATS-modified ␤-tubulin peptide revealed an increase in mass by 72.1 Da, corresponding to the mass of a fragment molecule derived from DATS, S-allylmercaptocysteine, i.e. conversion of a pro-  tein sulfhydryl group (protein-SH) to an oxidized form by DATS (protein-SS-allyl). In the ␤-tubulin, evidence of cysteine residue modification was proved by the detection of peptide 3 Fig. 6). Cys-12␤ and Cys-354␤ were identified as the residues that increased the peptide mass of 72.1 Da, suggesting that DATS-modified residues were included in the cysteinecontaining peptide fragments. On the other hand, DATS-related modification of cysteine residues other than Cys-12␤ or Cys-354␤ could not be detected in either ␣or ␤-tubulin (data not shown). These data suggest that among the tubulin structure, only 2 amino acid residues, Cys-12␤ and Cys-354␤, are the specific amino acid residues oxidized by DATS. To examine the specificity further, we also measured the number of cysteine residues in the tubulin incubated with DATS by titrating the -SH group with 5,5Ј-dithiobis-2-nitrobenzoic acid. The number of -SH groups present in the tubulin was decreased from 20 to 18.5 by the incubation of tubulin with 10 M DATS for 20 min. This result also supports the oxidation of two sulfhydryl groups by DATS at Cys-12␤ and Cys-354␤ as could be seen by mass spectrometry.
Antitumor Activity of DATS in the Xenograft Model Mice-As described so far, DATS is an effective anticancer component in the cell culture system in vitro. To verify the antitumor activity of DATS in vivo, we employed a HCT-15 xenograft mice model (Fig. 7). The growth of HCT-15 xenografts was significantly reduced in mice administered DATS in comparison with tumor growth in the vehicle-administered control mice. The mean tumor size measured at 27 days after the transplantation of the xenograft into the control mice was 1720 mm 3 , and that of the DATS-administered mice was as small as 518 mm 3 , demonstrating a 70% reduction in tumor size by DATS administration. DATS did not cause any change in body weight or any obvious side effects in this experiment. The histological examination of the tumor sections stained with hematoxylin and eosin revealed more necrotic region in DATS-treated mice than vehicle-treated mice (Fig. 7B).

DISCUSSION
We initially examined the effects of DAS, DADS, and DATS on the growth of human colon cancer cells to elucidate the relationship between the structure and function of allyl sulfides. The reason why DATS is more effective than DADS or DAS remains unclear, but because DATS is known to react with the sulfhydryl group of cysteine (20), modification of cysteine would be crucial for inhibiting cell growth. We found that incubation of cysteine with DATS under the physical pH caused oxidation of the sulfhydryl group of cysteine; however, that with DAS or DADS caused no such oxidation (data not shown). Taken together, our results suggest that the number of sulfur atoms in allyl sulfides is an important factor for exhibiting their antiproliferative activity and that the biological activities of allyl sulfides might be dependent on the reactivity of sulfur atoms with cellular components such as cysteine.
To elucidate the antiproliferative mechanisms of DATS, we next examined the effect of DATS on cell cycle progression as well as on apoptotic cell death. DATS caused cell cycle arrest at the M phase, sustained cyclin B1 expression, and induced apoptotic cell death in HCT-15 and DLD-1 cells (Figs. 3 and 4). Immunofluorescence microscopy using anti-␤-tubulin antibody clearly showed that DATS inhibited the spindle formation (Fig. 5C). The mitotic arrest caused by microtubule-interfering agents is earlier found to precede apoptotic cell death, and the hypothesis that cell cycle arrest at mitosis is the primary signal to induce apoptosis has been widely accepted (21)(22)(23). Actually DATS caused mitotic arrest prior to the induction of apoptosis in HCT-15 cells. Apoptosis induced by antimitotic agents is associated with alterations in a variety of cellular signaling pathways (24). Although the process by which microtubule-interfering agents induce apoptosis is poorly understood, Bcl-2 family proteins play central roles in the regulation of mitochondrial membrane permeability and in the induction of apoptosis (25). A variety of different kinases has been implicated in the Bcl-2 phosphorylation caused by antimitotic agents, including c-Jun N-terminal kinase, c-Raf, extracellular signal-regulated kinases, and cyclin-dependent kinase 1 (26 -29). We found that c-Jun N-terminal kinase inhibitor (SP600125) blocked the induction of apoptotic cell death by DATS, although it did not block mitotic arrest (data not shown). Further experiments will be required to address the cellular signaling pathways causing mitotic arrest and apoptosis by DATS.
We have also showed that the direct oxidative modification of tubulin at Cys-12␤ and Cys-354␤ by DATS (Fig. 6). This is assumed to be a cause for the disruption of microtubule network formation. Microtu- FIGURE 5. Effect of DATS on the microtubule network formation. A and B, assembly of phosphocellulose-purified tubulin was measured by conducting a turbidity assay. Tubulin (1.5 mg/ml) was mixed with DAS, DADS, DATS, colcemid, or paclitaxel at 4°C and then warmed at 37°C to initiate the polymerization. The increase in absorbance was monitored at 340 nm. C, the effect of DATS (10 M), colcemid (100 ng/ml), or paclitacel (50 nM) on the microtubule network formation in DLD-1 cells was assessed by the immunofluorescence method using anti-␤-tubulin antibody (green). The nucleus was counterstained with propidium iodide (magenta). Scale bar, 20 m. DECEMBER 16, 2005 • VOLUME 280 • NUMBER 50 bules are highly dynamic polymers that are responsible for the accurate chromosome segregation during mitosis through the formation of the bipolar mitotic spindle. Thus, drugs that disrupt microtubule network formation have been applied for the treatment of malignant tumors (15,16). Each microtubule-interacting agent has its own putative binding site in microtubules, e.g. colchicine-binding site, Vinca alkaloid-binding site, paclitaxel-binding site, and other unknown sites (30). By using mass spectrometry, we demonstrated that DATS can oxidize the sulfhydryl group of tubulin (Cys-12␤ and Cys-354␤) to disulfide (formation of protein-SS-allyl). Tubulin has 20 cysteine residues: ␣and ␤-tubulin containing 12 and 8 cysteine residues, respectively (31,32). Cys-354␤ is near the colchicine-binding site (33). Vinca alkaloids also bind with a domain so called "Vinca domain" containing Cys-12␤, which is thought to be located close to the exchangeable GTP-binding site (34). Gupta et al. (35) recently demonstrated by a mutagenesis study using Saccharomyces cerevisiae that Cys-12␤ and Cys-354␤ residues play important roles in maintaining the structure and function of tubulin. Taken together, oxidative modification of Cys-12␤ and Cys-354␤ by DATS causes the dysfunction of tubulin.

DATS Induces Apoptosis by the Modification of ␤-Tubulin
The oxidative modification of cysteine residues by DATS (formation of protein-SS-allyl) is thought to be reversible, because the inhibited spindle formation by DATS was attenuated at least within 16 h of culture after the challenging of DATS (data not shown). In fact the cells arrested by DATS at the M phase went into the next G 1 phase and underwent apoptosis (Fig. 3A); thus DATS would act to delay the cell cycle progression at the M phase (Fig. 4). Protein sulfhydryl groups are known to be easily modified by S-glutathionylation (formation of pro-tein-SSG), which is a reversible oxidation, to form a protein disulfide bond (protein-SS-protein) (36,37). Post-translational modification of protein by oxidative stress can regulate protein function in the same manner as phosphorylation (38). Chaperones, cytoskeletal proteins, cell cycle regulators, signal transduction proteins, and metabolic or redox enzymes are also regulated by oxidative modification (36,39,40). A cytoskeletal protein, tubulin, can also be S-glutathionylated (41). The intracellular thiol homeostasis is maintained by the thioredoxin and glutaredoxin systems, which utilize NADH as reducing equivalents to reduce proteins (42). Thus, oxidative modification by DATS forming the protein-SS-allyl may be restored by these redoxins and glutaredoxin systems as well as S-glutathionylation (protein-SSG), and the disulfide bond formation between the proteins may be possible (protein-SS-protein).
Based on the findings from the studies in vitro, we also examined the effect of DATS on the tumor growth in mice as a xenograft model in vivo. DATS potently reduced the tumor size in comparison with vehicle-administered control mice; a 70% reduction in the tumor size was observed. These results strongly suggest that DATS suppresses the tumor cell growth even in vivo by the mechanisms observed in vitro.
In summary, we demonstrated for the first time that DATS, one of the phytochemicals derived from garlic, bound to specific cysteine residues in ␤-tubulin molecule to form S-allylmercaptocysteine and that this could be the sole cause of cell cycle arrest and successive apoptosis with the activation of caspase-3. In other words, this is the first finding that a garlic-derived anticarcinogenic sulfide binds chemically with one of the most important proteins for cell growth. We also demonstrated that DATS inhibited significantly the growth of human colon carcinoma cells in nude mice in vivo. Garlic is widely served as a unique spicy vegetable around the world, and several lines of evidence, obtained by both laboratory and epidemiological research, have proven the anticancer effect of garlic (43,44). DATS is responsible, at least in part, for the effect, and it might thus be a lead compound for designing novel anticancer drugs.