Discovery of a Small Molecule That Inhibits Cell Division by Blocking FtsZ, a Novel Therapeutic Target of Antibiotics*

The emergence of bacterial resistance to antibiotics is a major health problem and, therefore, it is critical to develop new antibiotics with novel modes of action. FtsZ, a tubulin-like GTPase, plays an essential role in bacterial cell division, and its homologs are present in almost all eubacteria and archaea. During cell division, FtsZ forms polymers in the presence of GTP that recruit other division proteins to make the cell division apparatus. Therefore, inhibition of FtsZ polymerization will prevent cells from dividing, leading to cell death. Using a fluorescent FtsZ polymerization assay, the screening of >100,000 extracts of microbial fermentation broths and plants followed by fractionation led to the identification of viriditoxin, which blocked FtsZ polymerization with an IC50 of 8.2 μg/ml and concomitant GTPase inhibition with an IC50 of 7.0 μg/ml. That the mode of antibacterial action of viriditoxin is via inhibition of FtsZ was confirmed by the observation of its effects on cell morphology, macromolecular synthesis, DNA-damage response, and increased minimum inhibitory concentration as a result of an increase in the expression of the FtsZ protein. Viriditoxin exhibited broad-spectrum antibacterial activity against clinically relevant Gram-positive pathogens, including methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococci, without affecting the viability of eukaryotic cells.

The emergence of bacterial resistance to antibiotics is a major health problem and, therefore, it is critical to develop new antibiotics with novel modes of action. FtsZ, a tubulin-like GTPase, plays an essential role in bacterial cell division, and its homologs are present in almost all eubacteria and archaea. During cell division, FtsZ forms polymers in the presence of GTP that recruit other division proteins to make the cell division apparatus. Therefore, inhibition of FtsZ polymerization will prevent cells from dividing, leading to cell death. Using a fluorescent FtsZ polymerization assay, the screening of >100,000 extracts of microbial fermentation broths and plants followed by fractionation led to the identification of viriditoxin, which blocked FtsZ polymerization with an IC 50 of 8.2 g/ml and concomitant GTPase inhibition with an IC 50 of 7.0 g/ml.

That the mode of antibacterial action of viriditoxin is via inhibition of FtsZ was confirmed by the observation of its effects on cell morphology, macromolecular synthesis, DNA-damage response, and increased minimum inhibitory concentration as a result of an increase in the expression of the FtsZ protein. Viriditoxin exhibited broad-spectrum antibacterial activity against clinically relevant Gram-positive pathogens, including methicillin-resistant Staphylococcus aureus and vancomycin-resistant Enterococci, without affecting the viability of eukaryotic cells.
Bacterial infection is a global health hazard. There are a number of very good clinically efficacious antibiotics in use today; however, because bacteria render almost all of these antibiotics less effective because of the development of resistance, new antibiotics with novel mechanisms of action are needed to overcome the emerging resistance problem. To date, there are ϳ150 -670 essential proteins that have been identified in bacteria as potential drug targets (1)(2)(3)(4)(5). Of those, only 15 are proven therapeutic targets. To overcome resistance, existing antibiotics have been modified extensively in an attempt to maintain activity against their targets. Because efficacious modification is becoming increasingly difficult, targeting new, unexplored essential proteins to identify new classes of small molecular weight compounds is critical for the discovery of new, non-cross resistant antibacterial agents. FtsZ is a GTPase and an essential protein for cell division with homologs in almost all species of eubacteria and archaea (6 -8). FtsZ shows limited sequence similarity to tubulin; however, the three-dimensional structure of FtsZ shows significant similarities with ␣ and ␤ tubulin (9). FtsZ is the most abundant of all bacterial cell division proteins, totalling ϳ10,000 -20,000 copies per single bacterium (10). During vegetative growth, bacterial cells divide into two identical copies. The first stage in the cell replication process is localization of the FtsZ protein at the site of cell division where self-polymerization occurs. The polymerized FtsZ recruits other cell division proteins, including FtsA (11)(12)(13), ZipA (14,15), FtsK (16), FtsQ (17,18), FtsL (19,20), FtsW (21,22), FtsI (22,23), and FtsN (24), either by direct physical interaction or by secondary protein-protein interactions, leading to the formation of a Z-ring and the initiation of the complex process of septation. All of these cell division proteins are localized at mid-cell, working in concert to constrict the cell and resulting in cell division. Because FtsZ is essential for bacterial cell division and shows a high degree of similarity among bacterial species, it presents an excellent novel target for antibacterial drug discovery. In this study, we identified viriditoxin as an inhibitor of bacterial cell division and studied its mechanism of action and broad-spectrum antibacterial properties.

EXPERIMENTAL PROCEDURES
Isolation, Purification, and Identification of Viriditoxin-Aspergillus sp. (MF6890) was isolated from herbivore dung collected in Arizona. It was grown on a seed medium consisting of 4 g of yeast extract, 8 g of malt extract, 4 g of glucose, and 1.5 g of junlon (Nihon Junyaku Co. Ltd, Tokyo, Japan) in 1 liter of distilled water at pH 7.0. After 4 days, it was transferred to a vermiculite-based production medium consisting of (per liter) 150 g of glucose, 20 g of glycerol, 4 g of yeast extract, 1 g of sodium nitrate, 3 g of monosodium glutamate, 0.5 g of Na 2 HPO 4, 1 g of MgSO 4 .7H 2 O, and 8 g of CaCO 3 . The fermentation broth was extracted with methyl ethyl ketone and successively chromatographed on Sephadex LH 20 in methanol followed by reversed phase high pressure liquid chromatography affording 2 mg/liter of viriditoxin. The structure of viriditoxin was elucidated by 1 H, 13 C, and two-dimensional NMR methods and mass spectral studies.
Protein Purification-FtsZ and FtsZT65C were purified as described (25) with some modification. Briefly, BL21(DE3)pLysS (Novagen) was transformed with plasmid pET11a-ftsZwt or pET11a-ftsZT65C and induced by adding 1 mM IPTG 1 to mid-log culture (A 600 ϭ 0.5 ϳ 0.6). Cultures were harvested after 1 h at 37°C. To purify FtsZ and FtsZT65C, the cells were allowed to lyse for 30 min at 0°C in Buffer A (40 mM Tris-Cl, pH 8.0, 5 mM EDTA, 140 mM NaCl, and 10% sucrose) containing 1 mg/ml lysozyme and Complete protease inhibitor cocktail (Roche Applied Science). The lysed cells were then sonicated with four 30-s bursts using a microprobe. After centrifugation at 40,000 rpm in a 45 Ti rotor for 1 h at 4°C, the supernatant was adjusted to 10 mM MgCl 2 , solid ammonium sulfate was added (0.17 g/ml), and the cells * 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  were centrifuged for 1 h again. Ammonium sulfate pellets of FtsZwt (wild type) were resuspended in Buffer B (100 mM Tris-Cl, pH 7.4, 5 mM MgCl 2 , 80 mM KCl, and 10% glycerol) while ammonium sulfate pellets of FtsZT65C were suspended in buffer B with 1 mM reducing agent Tris-(2-carboxyethyl)phosphine hydrochloride (TCEP-HCl). FtsZwt or FtsZT65C was loaded separately onto a DEAE-Sepharose Fast Flow column (Amersham Biosciences) and eluted with a KCl gradient from 80 to 600 mM KCl. The fractions were analyzed by SDS-gel electrophoresis. The FtsZ-or FtsZT65C-containing fractions were pooled and concentrated using Amicon. The fluorescein-labeled FtsZT65C was prepared by incubating the purified FtsZT65C with 2 mM fluorescein iodoacetamide for 25 min in the same buffer on ice, and the free dye was removed by using a Sephadex G25 column. All proteins were aliquoted and stored at Ϫ80°C.
FtsZ Polymerization Assay-The assay was done as described previously (25) with a minor modification. Briefly, the NUNC 96-well Silent Screen filter plates were pre-washed with wash Buffer C (100 mM Tris-Cl, pH 7.4, 67.5 mM KCl, 1.5 mM magnesium acetate, and 0.1% Tween.) followed by centrifugation at 3,000 rpm at room temperature for 15 min in a Beckman tabletop centrifuge. The fluorescence-labeled FtsZT65C (FtsZT65C-fluorescein) was pre-centrifuged at 100,000 ϫ g for 20 min at 4°C to removes any pre-existing polymers. The FtsZ polymerization assay was done by mixing 1 M fluorescence-labeled FtsZT65C with a serial dilution of inhibitors in 200 l of Buffer D (50 mM Tris pH 7.4, 34 mM KCl, 0.75 mM magnesium acetate, 2.5 mM CaCl 2 , 0.5 mM GTP, and 2% Me 2 SO) and initiated with 50 g/ml DEAEdextran. After incubation at 37°C for 15 min, the filter plates were centrifuged at 3,000 rpm for 15 min, and the fluorescence units were read on the Tecan SpectraFluor with 485/535-nm filters. The range of reading was between 3,000 and 40,000 fluorescence units. IC 50 was determined using Prism Graph.
GTPase Assay-GTPase activity was measured and analyzed as described (25, 26) with a small modification. Briefly, this assay was performed at 37°C in a 50-l buffer containing 50 mM Tris, pH 7.4, 34 mM KCl, 0.75 mM magnesium acetate, 2.5 mM CaCl 2 , 50 g/ml DEAEdextran, 2% Me 2 SO, GTP, and pre-centrifuged FtsZ or FtsZT65C-fluorescein. The concentrations of GTP and the proteins as well as the time of the reaction are indicated in the legends to Figs. 3-5. Assays were stopped by the addition of activated charcoal, and the supernatant was counted.
Minimum Inhibitory Concentration (MIC)-The MIC against each of the strains was determined according to the National Committee for Clinical Laboratory Standards guidelines. MIC is defined as the lowest concentration of antibiotic inhibiting visible growth.
Whole Cell-labeling Assay-The effect of viriditoxin on macromolecular synthesis was performed as described (27).
MIC Assay with FtsZ Expression-A permeable (envAϪ) BL21(DE3)pLysS-derivative Escherichia coli strain was constructed using standard P1 transduction methodology (28), utilizing a Tn10 transposon linked to the envA1 gene (29). The cells were transformed with the plasmid pET11a-ftsZT65C. The MICs of viriditoxin against the transformed cells were assayed at increasing concentrations of IPTG in LB medium containing 25 g/ml of ampicillin and 5 g/ml chloramphenicol at 37°C for 20 h. All reagents were purchased from Sigma unless indicated otherwise.

RESULTS AND DISCUSSION
A high-throughput FtsZT65C-fluorescein polymerization assay has been developed (25) for identification of novel inhibitors. Using this assay, we screened Ͼ100,000 extracts of microbial fermentation broths and plants, getting a hit rate of 0.043%. All the active extracts were tested secondarily for filament morphology and also in an SOS induction assay (30) to further refine the quality of the actives providing only eight promising hits. Although many of these are still under active investigation, bioassay-guided purification of one of the most promising active extracts derived from an Aspergillus sp. led to the isolation of viriditoxin (Fig. 1), which was first reported in 1971 from Aspergillus viridinutans (31). At the time of original discovery, the structure of viriditoxin was incorrectly assigned and was reported to have an LD 50 of 2.8 mg/kg to mice (31). The structure was corrected in 1990 (32). The spectral data (NMR, mass spectroscopy, and UV) of the isolated compound was identical to the reported data of viriditoxin (32). However, a subsequent report in 1976 suggested that it showed a lower level of toxicity to pregnant mice at up to 3.5 mg/kg for 10 days (33). Interestingly, it was reported that it had little effect on prenatal survival or growth (33).
The activity of purified viriditoxin was first confirmed in the FtsZT65C-fluorescein polymerization assay. It inhibited the polymerization with an IC 50 average value of 8.2 g/ml (n ϭ 3) (Fig. 2).
To confirm that viriditoxin inhibits wild type FtsZ polymerization, we compared the GTPase activity of the wild type FtsZ directly with FtsZT65C-fluorescein. For this study, we used purified wild type FtsZ and FtsZT65C-fluorescein. The K m values for GTP of wild type FtsZ and FtsZT65C-fluorescein are similar, providing an average value of 19.4 M (n ϭ 2) and 19.7M (n ϭ 3), respectively. The V max of FtsZT65C-fluorescein is 1.15 mol of P i /mol of protein/min, whereas the V max of wild type FtsZ is 1.40 mol of P i /mol of protein/min (an 18% difference; Fig. 3). The K m of wild type FtsZ was comparable with data reported in previous publications (34), and the V max of wild type FtsZ was also comparable with published data (35), although the assay conditions were not identical. At a fixed concentration of 40 M GTP with 1 M wild type FtsZ or 1 M FtsZT65C-fluorescein, at 37°C, the t1 ⁄2 values of steady-state GTP hydrolysis were 11.7 and 15.9 min, respectively (Fig. 4). Because GDP/GTP exchange (35) is not rate-limiting, the apparent rate constants of GTP hydrolysis of both wild type FtsZ and FtsZT65C-fluorescein are apparently 0.059 and 0.043 min Ϫ1 , respectively, further indicating that both proteins have similar GTPase properties.
Because the polymerization of FtsZ is GTP-dependent and the GTPase activity of FtsZ is activated by self-polymerization (34), we first measured GTP hydrolysis under conditions in which the polymerization of FtsZT65C-fluorescein was not detectable (0.2M FtsZT65C-fluorescein; Fig. 4, ‚). GTP hydrolysis in 120 min was only 2 mol per mol of protein. Second, we tested the effect of viriditoxin on GTPase activity with prepolymerized FtsZT65C-fluorescein (Fig. 4, f). In the presence of 200 g/ml viriditoxin, GTP hydrolysis was 2 mol per mol of protein in 120 min, consistent with the near total inhibition of polymerization by this level of viriditoxin. In contrast, the GTP hydrolysis of both wild type FtsZ and FtsZT65C-fluorescein under identical conditions but without the inhibitor was 32.7 and 33.7 mol per mol of protein, respectively (Fig. 4, E and q). Titration of viriditoxin into the GTP hydrolysis assay (Fig. 5) yielded an average IC 50 of 7.0 g/ml (n ϭ 2).
To measure directly the effect of viriditoxin on cell division, a morphology assay was employed (36). Highly permeable E. coli (MB5431) cells lacking the SulA protein, the natural inhibitor of FtsZ polymerization and GTPase that is normally induced by the SOS response leading to filamentation (36) toxin-treated cells formed filaments at 50 (Fig. 6A) as well as at 12.5, 25, and 100 g/ml (data not shown). Without viriditoxin treatment, cells in mid-log did not form filaments (Fig. 6B). For comparison, we tested a strain in which SulA production is under lac promoter control. The strain, HMS174 (DE3, plysS) containing pET11-SulA, was grown to log phase and then induced with IPTG for 75 min, resulting in morphological behavior similar to what was seen with viriditoxin-treated cells (Fig. 6C). Without induction by IPTG, the cells did not form filaments (Fig. 6D).
To confirm that viriditoxin interacts with FtsZ in bacteria, we performed a MIC assay with the permeable E. coli-bearing pET11a-ftsZT65C. First, we tested the lethal dose of IPTG induction without the addition of viriditoxin and found that 8 M g/ml, indicating that penetration plays a role in the insensitivity of the compound of Gram-negative bacterial strains. In comparison, it did not show activity against Candida albicans (MIC Ͼ64 g/ml) and human HeLa cells (MTS, IC 50 Ͼ66.7 g/ml), indicating a selectivity for prokaryotic over eukaryotic cells.
Our in vitro results indicated that viriditoxin inhibits FtsZ polymerization and GTPase activity. To our knowledge, this is the first small organic non-peptide molecule (molecular mass, 662 Da) shown to act as a cell division inhibitor by blocking FtsZ polymerization. Because the three-dimensional structure of FtsZ is strikingly similar to tubulin, we titrated the known tubulin inhibitors (albendazole, colchicine, nocodazole, paclitaxel, 3-methoxybenzamide, and thiabendazole) in FtsZ polymerization and GTPase assays. Up to 167 g/ml, none of these compounds showed any inhibition of FtsZ polymerization or GTPase activity in vitro (data not shown), thus differentiating the FtsZ from tubulin and the viriditoxin from classical tubulin FIG. 6. Effect of viriditoxin on cell division. A and B, an overnight culture of the SOS negative, sulA Ϫ K12-derivative E. coli strain MB5431 (sulA::MUD, uvrA155, envA, Tn10, lexA102) was diluted in fresh LB broth (A 600 ϭ 0.1) and cultured at 37°C for 1 h with shaking at 200 rpm (A 600 ϳ 03). A serial dilution of viriditoxin was added to the culture, and the cells were further cultured at 37°C with constant shaking for 75 min. For panel A, treatment was with 50 g/ml of viriditoxin in 2% Me 2 SO; for panel B, treatment was with 2% Me 2 SO (control). C and D, strain HMS174 (DE3, pLysS; Novagen) containing pET11-SulA (36). In panel C, mid-log (A 600 ϭ 0.6) cultured cells were induced by the addition of 1 mM IPTG for 75 min. In panel D, mid-log culture cells were used as control. inhibitors. The results obtained from the whole cell experiments, particularly the filamentation experiment with the E. coli (SOSϪ, SulAϪ) strain, showed that filaments were not formed because of DNA damage and confirmed the target selectivity of viriditoxin. Viriditoxin was not active (data not shown) in a modified SOS chromo test (30), further corroborating that it was not involved in DNA damage. The increased MIC by induction of FtsZ expression indeed gives solid evidence that viriditoxin interacts with FtsZ in the cell. In whole cell labeling experiments using wild type S. aureus MB2985 (MIC ϭ 3.25 g/ml) and the permeable E. coli MB5431 (MIC ϭ 25 g/ml), viriditoxin did not block DNA, RNA, protein, fatty acid, and cell wall synthesis (for all macromolecular syntheses in both strains the IC 50 was Ͼ100 g/ml; data not shown), further supporting the activity as a result of a new mode of action. The broad-spectrum antibacterial activity of this compound, including methicillin-resistant S. aureus, vancomycinresistant Enterococci, and other antibiotic resistant strains, would indicate that the activity of this compound is due to blockage of a new target, i.e. FtsZ. Furthermore, although viriditoxin was identified initially by its activity against the purified E. coli (Gram-negative) FtsZ protein, the in vivo activity of viriditoxin against many Gram-positive bacteria indicates a high functional conservation of the target in these clinically important species. Presumably, the structure of the FtsZ molecule is very highly conserved in the pocket that is engaged by viriditoxin, and this may limit the ability to develop resistance in the target. Although viriditoxin may not be a drug itself, it could be a lead for drug discovery as well as an important tool for biological studies in the field of cell division.