Mechanism of Action of Cryptophycin INTERACTION WITH THE Vinca ALKALOID DOMAIN OF TUBULIN*

Cryptophycin is a potent antitumor agent that de- pletes microtubules in intact cells, including cells with the multidrug resistance phenotype. To determine the mechanism of action of cryptophycin, its effects on tubulin function in vitro were analyzed. Cryptophycin reduced the in vitro polymerization of bovine brain microtubules by 50% at a drug:tubulin ratio of 0.1. Cryptophycin did not alter the critical concentration of tubulin required for polymerization, but instead caused substoichiometric reductions in the amount of tubulin that was competent for assembly. Consistent with its persistent effects on intact cells, cryptophycin-treated microtubule protein remained polymerization-defective even after cryptophycin was reduced to sub-inhibitory concentrations. The effects of cryptophycin were not due to denaturation of tubulin and were associated with the accumulation of rings of microtubule protein. The site of cryptophycin interaction with tubulin was examined using functional and competitive binding as- says. Cryptophycin blocked the formation of vinblas-tine-tubulin paracrystals in intact cells and suppressed vinblastine-induced tubulin aggregation in vitro . Cryptophycin inhibited the binding of [ 3 H]vinblastine and the hydrolysis of [ (cid:103) - 32 P]GTP by isolated tubulin, but did not block the binding of colchicine. These results indi- cate that cryptophycin disrupts the Vinca alkaloid site of tubulin; however, the molecular details of this inter- action are distinct from those of other antimitotic drugs.

Microtubules are extraordinarily dynamic assemblies, which are involved in many cellular activities including maintenance of cell structure and regulation of cell motility, membrane transport processes and cell proliferation (reviewed in Ref. 1). The process of continuous addition and loss of tubulin dimers from microtubule ends, termed dynamic instability (2), allows rapid remodeling of these structures, but also makes them susceptible to chemical agents that disrupt microtubule structure and function. Inhibition of microtubule dynamics in the mitotic spindle is thought to be the basis for the anticancer activities of drugs that promote the depolymerization of microtubules, e.g. Vinca alkaloids, or that stabilize microtubules, e.g. taxanes (3)(4)(5)(6)(7). These compounds are widely used clinically; however, the ability of tumor cells to develop resistance to these, and most other, natural product anticancer drugs often limits their ultimate efficacy (reviewed in Refs. 8 -10).
We have recently described the ability of a cyanobacterial cytotoxic macrolide, termed cryptophycin, to induce the depletion of microtubules in intact cells (11). Importantly, cryptophycin is able to circumvent a common form of multiple drug resistance, i.e. P-glycoprotein-mediated efflux of natural product anticancer drugs. This confers a significant theoretical advantage to this compound over currently utilized antimitotic drugs in that tumors should be less able to develop resistance to cryptophycin. Cryptophycin has demonstrated excellent in vivo activity against several types of tumor xenografts, including those of cells that are poorly inhibited by Vinca alkaloids (12). Because of the potential usefulness of cryptophycin as an anticancer drug, we have now examined the molecular mechanism of its antimicrotubule actions. These studies have revealed that cryptophycin interacts with the Vinca alkaloid binding domain of tubulin; however, a number of unique molecular details of cryptophycin binding have been identified.

EXPERIMENTAL PROCEDURES
Materials-Vinblastine, colchicine, podophyllotoxin, and antibodies against ␤-tubulin (T-4026) were obtained from Sigma. Rhizoxin was obtained from the Drug Synthesis and Chemistry Branch, NCI, National Institutes of Health. Cryptophycin was provided by Dr. R. E. Moore of the University of Hawaii, or synthesized by methods similar to those of Barrow et al. (13) and Kobayashi et al. (14). It should be noted that the original stereochemical analysis of cryptophycin (12) resulted in an incorrect assignment at the chloro-O-methyltyrosine moiety (13). The proper structure for cryptophycin is indicated in Fig. 1. Isolation of Microtubule Protein and Tubulin-Bovine brains were obtained from a local slaughterhouse and kept at 4°C for approximately 1 h before cleaning and homogenization. Microtubule protein (MTP) 1 was isolated by two cycles of polymerization-depolymerization as described by Vallee (15) and consisted of approximately 75% tubulin and 25% microtubule-associated proteins (MAPs), as estimated by Coommassie Brilliant Blue staining of MTP after SDS-polyacrylamide gel electrophoresis (data not shown). For some studies, tubulin was purified from MTP by chromatography on DEAE-Sepharose using the procedure of Vallee (15). The concentration of purified tubulin was determined by its absorbance at 275 nm (⑀ ϭ 1.07 ml/(mg ϫ cm)) (16). MTP and purified tubulin were stored in liquid nitrogen in their polymerized forms.
Polymerization Assays-In the standard polymerization assay (17), samples (0.25 ml) containing 2.5 mg of MTP/ml (20 M tubulin) in MTP polymerizing buffer (0.1 M MES, pH 6.4, containing 0.5 mM MgCl 2 ) were incubated with drug at 4°C for 15 min before the addition of 0.5 mM GTP. The samples were then rapidly warmed to 37°C in a waterjacketed cuvette holder of a Pharmacia Ultraspec III spectrophotometer, and the absorbance at 350 nm was monitored for approximately 30 min. For studies with purified tubulin, 0.25-ml samples containing 2 mg of tubulin/ml (20 M) in 1 M glutamate, pH 6.6, 0.5 mM MgCl 2 , 4% dimethyl sulfoxide, and drug were incubated at 4°C for 15 min. Tubulin polymerization was initiated and monitored as indicated above. The following variations of this polymerization assay were used in certain studies.
1) To examine the effects of drugs on microtubule depolymerization, MTP was polymerized in the absence or presence of 10 M paclitaxel for 30 min under the conditions described above. Cryptophycin or vinblastine was then added, and the A 350 was monitored for approximately 30 min.
2) To examine the effects of drugs on microtubule morphology, MTP samples were incubated with drug under polymerizing conditions for 30 min. The samples were then fixed with 0.2% glutaraldehyde, absorbed onto carbon-coated grids, and stained with 0.5% uranyl acetate for electron microscopy.
3) To test the effects of cryptophycin on the critical concentration of tubulin, the amount of MTP was varied from 0.25 to 5 mg/ml (2-40 M tubulin) and cryptophycin concentrations were varied from 0 to 10 M. Polymerization was initiated and monitored as indicated above. 4) To test the reversibility of the inhibitory effect of cryptophycin, MTP was incubated with 5 M cryptophycin for 30 min at room temperature in MTP polymerizing buffer (without GTP). The samples were then diluted 10-fold with cold MTP polymerizing buffer and concentrated 10 -80-fold by ultrafiltration using Centricon-10 centrifugal concentrators (Amicon, Inc., Beverly, MA), which contain M r 10,000 cut-off filters. After measurement of the volume of the concentrate, the samples were diluted to their original volume (0.25 ml) so that the MTP was restored to its original concentration and cryptophycin was reduced to Յ0.5 M. A 0.2-ml aliquot was then combined with 0.5 mM GTP and warmed to 37°C, and polymerization was monitored as indicated above. The remainder of the sample was combined with 9 volumes of cold EtOH, and precipitated protein was pelleted by centrifugation at 15,000 ϫ g for 10 min. The supernatant was serially diluted and tested for cytotoxicity toward MCF-7 cells as described previously (11). The actual amount of cryptophycin in each centrifuged sample was determined by comparison with the cytotoxicities of cryptophycin standards.
Ligand Binding and GTPase Assays-The ability of cryptophycin to interact with tubulin was also examined by determining its ability to modulate the binding of colchicine, [ 3 H]vinblastine, [ 3 H]GTP, and bisANS, and the hydrolysis of [␥-32 P]GTP according to the following protocols.
1) In the colchicine binding assay (18), triplicate samples of bovine brain MTP (10 M tubulin) were incubated with cryptophycin, podophyllotoxin or vinblastine in 0.1 M MES, pH 6.6, containing 1 mM MgCl 2 , 1 mM GTP, 100 mM glucose 1-phosphate (19), for 5 min before the addition of 10 M colchicine. After 30 min, the fluorescence of the MTP mixture were determined using an excitation wavelength of 357 nm and an emission wavelength of 435 nm. Increased fluorescence of colchicine is observed upon its binding to tubulin (18). Samples without colchicine or without MTP were utilized as blanks.
2) In the [ 3 H]vinblastine binding assay (20), triplicate samples of MTP (2 M tubulin) were incubated in 0.1 M MES, pH 6.9, containing 0.5 mM MgCl 2 , 2% dimethyl sulfoxide, and 5 M [ 3 H]vinblastine (0.1 Ci) with several concentrations of test drugs for 30 min at 21°C. The samples (0.15 ml) were then applied to 0.5-ml columns of Sephadex G-50 equilibrated with incubation buffer without vinblastine and centrifuged at 100 ϫ g for 2 min. Additional buffer (0.3 ml) was then added to each column, and the samples were centrifuged again. An aliquot of the filtrate (0. 25 (21), triplicate samples of 20 M purified tubulin were incubated in 1 M glutamate, pH 6.9, containing 0.5 mM MgCl 2 , 2% dimethyl sulfoxide, and 0.1 mM [␥-32 P]GTP (0.1 Ci) (22) with several concentrations of test drugs for 5 min at 37°C. Reactions were terminated by the addition of 1 ml of 300 mM perchloric acid containing 10 mg of activated charcoal (23). Samples were then incubated at room temperature for 5 min to allow adsorption of the nucleotide to the charcoal, followed by centrifugation at 3,000 ϫ g for 5 min. An aliquot (0.25 ml) of the supernatant was removed and analyzed by liquid scintillation counting to determine the amount of released 32 PO 4 . Blank samples did not contain MTP, or were stopped immediately after the addition of [␥-32 P]GTP, and their counts were subtracted from test samples. 5) To quantify the rates of tubulin denaturation (24), purified tubulin (5 M) was incubated at 37°C in 100 mM MES, pH 6.4, containing 0.1 mM EDTA, 1 mM EGTA, 1 mM GTP, 0.5 mM MgCl 2 , and 1 mM ␤-mercaptoethanol in the absence or presence of cryptophycin. At 30-min intervals, samples were combined with 25 M bisANS and the fluorescence was determined using an excitation wavelength of 385 nm and an emission wavelength of 490 nm (24). Samples without bisANS or without tubulin were utilized as blanks.
Tubulin Immunofluorescence Assay-Human ovarian carcinoma cells (SKOV3) were grown on glass coverslips in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum as described previously (25). The cells were treated with the indicated concentration of cryptophycin, colchicine, or rhizoxin for 3 h (which is sufficient to cause complete loss of microtubules) and then exposed to 5 M vinblastine for 18 h. Cells were fixed with cold methanol and stained with an anti-␤tubulin monoclonal antibody, followed by fluorescein-conjugated antimouse IgG as described previously (11). Images were collected by confocal microscopy (Bio-Rad MRC 6000) and processed using Voxel View™ software.
Photolabeling of Tubulin with [ 3 H]Colchicine-Similar to the procedure of Uppuluri et al. (26), MTP (10 M tubulin) was incubated in the dark for 15 min with 2 M [ 3 H]colchicine (2 Ci/assay) in the buffer utilized for the colchicine binding experiments in flat-bottomed polystyrene microtiter plates. The plates were then exposed to 365 nm light at 1 milliwatt/cm 2 for 10 min. MTP samples were then dissolved in Tris/ glycine loading buffer and were electrophoresed on 10 -20% gradient gels in the presence of 0.1% SDS. The gel was then soaked in Amplify (Amersham Corp.), dried, and exposed to Hyperfilm-MP x-ray film for up to 14 days, as necessary. Band intensities on the fluorograms were quantified using the program NIH-Image.
Topoisomerase Inhibition Assays-To test the effects of drugs on topoisomerase II-mediated cleavage of supercoiled pBR322, 0.6 g of the plasmid was incubated alone or in the presence of 20 units of calf thymus topoisomerase II (U. S. Biochemical Corp.) for 40 min at 30°C. Reactions were conducted in the presence of EtOH (as a solvent control), 10 M amsacrine, 10 M cryptophycin, or 50 M camptothecin. Samples were then electrophoresed on a 1% agarose gel (Tris acetate buffer, pH 7.9) and stained with ethidium bromide for visualization of DNA.
Additionally, the K ϩ /SDS precipitation assay was used to test the effects of drugs on topoisomerases in intact cells (27). Briefly, SKOV3 cells were grown for 24 h in the presence of [ 3 H]thymidine (1 Ci/ml) in 24-well tissue culture plates. At that time, the medium was replaced with unlabeled medium containing 10% fetal bovine serum and cells were exposed to multiple doses of test drugs for 45 min. The cells were then lysed with 1.25% SDS, and protein-complexed DNA was precipitated by the addition of 325 mM KCl. Pellets were washed twice, and the amount of precipitated 3 H was determined by liquid scintillation counting.

Effects of Cryptophycin on Microtubule Polymerization in
Vitro-The effects of equal concentrations of colchicine, vinblastine and cryptophycin on the in vitro assembly of MTP are demonstrated in Fig. 2. Microtubule assembly from MTP was inhibited by approximately 50% by 5 M colchicine, while this concentration of vinblastine and cryptophycin strongly reduced microtubule polymerization. The initial rate of microtubule assembly was unaffected by 5 M colchicine, most likely reflecting the relatively slow binding of this drug (4). In contrast, both vinblastine and cryptophycin suppressed the rate as well as the Substoichiometric doses of cryptophycin (with respect to tubulin) suppressed microtubule assembly (Fig. 3A), such that polymerization was decreased by approximately 50% by a cryptophycin:tubulin ratio of 0.1. Cryptophycin concentrations of 0.5 M and below did not affect microtubule assembly. In contrast with vinblastine (4), high doses of cryptophycin (up to at least 40 M) did not promote the formation of aberrant tubulin aggregations that result in increased light scattering (data not shown). Cryptophycin also suppressed the ability of purified tubulin to assemble into microtubules (Fig. 3B), indicating that the molecular target of the compound is tubulin rather than a MAP. To examine the polymerization of tubulin in the absence of MAPs, it is necessary to utilize reaction conditions which strongly promote microtubule assembly, i.e. 1 M glutamate, pH 6.6, containing 4% dimethyl sulfoxide (4). The reduced effect of cryptophycin on purified tubulin most likely reflects the difficulty in blocking assembly under these conditions. Experiments in which the concentrations of both tubulin and cryptophycin were varied indicated that the compound altered the percentage of tubulin which was competent for assembly, but did not alter the critical concentration of tubulin required for assembly (Fig. 4). These effects of cryptophycin are clearly substoichiometric with respect to tubulin, e.g. increasing tubulin from 10 to 40 M allowed only a small increase in assembly in the presence of 5 M cryptophycin. Electron microscopy revealed that MTP incubated with cryptophycin accumulated as rings (Fig. 5). The scattered aggregates of protein do not have structural similarity with spirals formed in the presence of vinblastine (Ref. 28 and data not shown).
Pretreatment of intact cells with paclitaxel prevents microtubule depletion upon exposure to cryptophycin or vinblastine (11). To test the effects of paclitaxel in vitro, microtubules were assembled in the absence or presence of 10 M paclitaxel and then exposed to cryptophycin or vinblastine. As demonstrated in Fig. 6, addition of cryptophycin to microtubules assembled in the absence of paclitaxel led to rapid, partial disassembly, manifested as a decrease in turbidity. In contrast, vinblastine was ineffective in reducing the state of microtubule assembly. Paclitaxel enhanced microtubule assembly, in agreement with previous studies (29), and markedly reduced the ability of cryptophycin to induce disassembly.
Treatment of cells with cryptophycin results in irreversible depletion of microtubules (11). To test the reversibility of the in vitro effects of the compound, MTP was treated with 5 M cryptophycin, concentrated 10-or 30-fold by ultrafiltration, and then diluted to the original volume. Residual concentrations of cryptophycin in the MTP sample were quantified by a bioassay for cytotoxicity and were found to amount to 0. 16  tophycin was unable to assemble (Fig. 7), even when residual levels of cryptophycin were reduced to 0.085 M. This concentration of cryptophycin has no effect on MTP assembly, indicating that treatment with cryptophycin caused persistent inhibition of assembly.
Incubation of tubulin at 37°C results in time-dependent denaturation that can be monitored as increased binding of bisANS to hydrophobic sites (24). Cryptophycin (5 M or greater) completely blocked the spontaneous increase in the ability of tubulin to bind bisANS (data not shown). This stabilization of tubulin supports the hypothesis that cryptophycin binds directly to tubulin and indicates that the persistence of the inhibitory effect of cryptophycin of polymerization is not due to nonspecific denaturation of tubulin.
Effects of Cryptophycin on the Drug Binding Domains of Tubulin-Interaction of cryptophycin with the colchicine and Vinca alkaloid domains of tubulin was assessed by determining its ability to compete with colchicine and [ 3 H]vinblastine binding, respectively. As indicated by Fig. 8A,  Colchicine binding to tubulin can be monitored by an increase in its fluorescence at 435 nm (18). Pretreatment of MTP with podophyllotoxin strongly reduced colchicine binding, whereas neither vinblastine nor cryptophycin altered the increased fluorescence of colchicine upon its interaction with tubulin (Fig. 8B)   To verify the ability of cryptophycin to modulate the Vinca binding domain of tubulin, its effects on the formation of vinblastine-tubulin paracrystals in intact cells were examined. Treatment of ovarian carcinoma (SKOV3) cells with 5 M vinblastine resulted in the depletion of microtubules and the formation of large paracrystals of vinblastine-tubulin (Fig. 9). Treatment of cells with 5 M colchicine, 5 M rhizoxin, or 25 nM cryptophycin depleted microtubules, resulting in diffuse cytoplasmic staining of ␤-tubulin. Addition of vinblastine to colchicine-treated cells resulted in the formation of paracrystals; however, no paracrystals were formed in cells pretreated with either rhizoxin or cryptophycin. Similarly, pretreatment of MTP in vitro with 10 M cryptophycin blocked the ability of high doses of vinblastine (up to at least 80 M) to cause tubulin aggregation (data not shown).
Lack of Effect of Cryptophycin on Topoisomerases-Since several antimitotic agents are cytotoxic through inhibition of topoisomerases, we directly tested the effects of cryptophycin on topoisomerases in vitro and in intact cells. In vitro cleavage of supercoiled plasmid pBR322 by topoisomerase II was blocked by 10 M amsacrine, whereas 10 M cryptophycin and 50 M camptothecin did not inhibit topoisomerase II activity (data not shown). Inhibition of topoisomerases in intact cells is associated with the accumulation of "cleavable complexes" consisting of drug, topoisomerase, and DNA which can be easily

DISCUSSION
Our initial descriptions of the actions of cryptophycin were limited to cellular studies in which this cyanobacterial natural product demonstrated extremely potent cytotoxicity, which was closely correlated with mitotic arrest, and pronounced, irreversible depletion of cellular microtubules (11). A major point of interest with this compound is its ability to kill cells that are resistant to other antimitotic drugs, including vinblastine and paclitaxel. Because of this, cryptophycin is an attractive agent for development as an anticancer drug. To further characterize the molecular actions of cryptophycin, we have now examined the effects of this compound on tubulin function in vitro.
Antimitotic compounds commonly disrupt mitotic spindle function by interfering with microtubule dynamics (30 -32), and this effect is reflected in their abilities to modulate the assembly of microtubules in vitro. As with other microtubuledepleting agents, cryptophycin strongly suppressed the ability of microtubules to assemble in vitro, indicating that the target of its action is contained in these preparations. Cryptophycin inhibited the assembly of purified tubulin, prevented tubulin denaturation, inhibited tubulin's ability to bind [ 3 H]vinblastine and to hydrolyze [␥-32 P]GTP. These results demonstrate that cryptophycin directly interacts with tubulin. Similar reduced efficacy for inhibition of the assembly of purified tubulin in comparison with assembly of microtubules from MTP has been observed with other antimitotic compounds (19, 33, 34) and most likely reflects the difficulty in preventing polymerization under conditions that strongly favor assembly, i.e. high glutamate and dimethyl sulfoxide concentrations (4).
Microtubules assembled in the presence of paclitaxel were stable upon treatment with cryptophycin, suggesting that cryptophycin binds to unpolymerized tubulin and prevents the growth of microtubules. Additionally, the effects of paclitaxel on alterations of microtubules in intact cells by cryptophycin (11) are fully mimicked by the MTP preparation, indicating that no other regulatory proteins mediate the actions of cryptophycin.
Interestingly, cryptophycin did not alter the critical concentration for microtubule assembly as has been reported for other tubulin-binding agents (35)(36)(37)(38). Instead, cryptophycin induced substoichiometric reductions in the maximum extent of assembly. This is consistent with an "end-poisoning" model in which addition of tubulin-cryptophycin complexes to the microtubule blocks further elongation (39,40). In support of this mechanism, electron microscopy indicated that the modest increase in turbidity of MTP incubated in the presence of cryptophycin is associated with the accumulation of tubulin rings (Fig. 3). These structures have been previously observed under conditions in which the formation of microtubules is prevented by low temperature or lack of GTP (41), and strikingly contrast with the effects of vinblastine and colchicine which induce the accumulation of spirals (28) and twisted ribbons (42), respectively.
It may be noted that microtubule depletion in intact cells can be accomplished by nanomolar doses of cryptophycin, while inhibition of microtubule polymerization in vitro requires micromolar levels of the compound. This apparent discrepancy is also observed with Vinca alkaloids (43), paclitaxel (32) and colchicine, 2 and is due to the high concentration of tubulin needed to measure microtubule assembly in vitro, i.e. 20 M. Consequently, effects mediated by stoichiometric binding of the agent to tubulin are observed only at micromolar concentrations. Additionally, other tubulin-binding drugs have been shown to concentrate Ͼ100-fold within cells, greatly increasing their effective levels at the target site. It is likely that this consideration also applies to cryptophycin. Recent work has demonstrated that microtubule dynamicity, i.e. addition and removal of tubulin dimers from microtubule ends, is considerably more sensitive to inhibition by antimitotic drugs than are alterations of total cellular microtubule mass (30 -32). In support of a similar mechanism for cryptophycin, dual-labeling studies have demonstrated that cryptophycin induces mitotic arrest and apoptosis in cells with intact cytoplasmic microtubules. 3 Since the time that tubulin was identified as the intracellular binding site of colchicine (44), a number of natural and synthetic compounds that alter microtubule structure and function have been characterized. Natural products that act as antimitotic agents due to depletion of cellular microtubules have been classified according to their abilities to bind to either the colchicine site or the vinblastine site of tubulin (4). While neither drug-binding site has been structurally characterized, some similarities among compounds interacting with these sites can be recognized. The classification of new compounds generally relies on study of their effects on the binding of colchicine and vinblastine. To date, compounds (and their analogues) have segregated quite well into these two categories. For example, Vinca alkaloids, rhizoxin (45), maytansinoids (46), phomopsins (47), dolastatins (48), halichondrins (33), ustiloxins (49), and spongistatin (50) interact with the "vinblastine" site, while colchicine, podophyllotoxins (51), steganacin (52), combrestatins (53), and curacin A (54) interact with the "colchicine" site.
Interaction of cryptophycin with the Vinca alkaloid binding site of tubulin is indicated by its ability to inhibit the binding of [ 3 H]vinblastine, as well as its ability to inhibit the formation of drug-tubulin paracrystals in cells or MTP treated with high doses of vinblastine. Attempts to use the DEAE-filter assay to measure [ 3 H]colchicine binding to tubulin (19) resulted in very low binding stoichiometry (Ͻ0.1), clearly indicating that the reaction had not reached equilibrium. This probably reflects the relatively low k on for colchicine interaction with tubulin (4,55) and the instability of the colchicine binding site of tubulin incubated at 37°C (4). Therefore, two independent assays were used to examine the interaction of cryptophycin with tubulin. These studies indicated that cryptophycin did not alter photolabeling of tubulin by [ 3 H]colchicine or alter increases in colchicine fluorescence upon its binding to tubulin (Fig. 8B).
The properties of previously known natural products which modulate the Vinca alkaloid domain of tubulin have recently been summarized by Luduena et al. (56) and indicate that cryptophycin has molecular effects that are unique. For example, vinblastine, phomopsin A, and dolastatin 10 enhance colchicine binding to tubulin, whereas cryptophycin is without effect. Tubulin decay in vitro is strongly inhibited by cryptophycin, vinblastine, phomopsin A, ustiloxin A, and dolastatin 10, but is increased or not affected by maytansine, rhizoxin, homohalichondrin B, and halichondrin B. Binding of GTP to tubulin is not affected by cryptophycin or vinblastine, but is inhibited by the other natural products. Cryptophycin and vinblastine have markedly different effects on tubulin structure both in vitro and in intact cells. Therefore, cryptophycin binding to tubulin is mechanistically different from previously described compounds and so should be useful in further defining drug binding domains on this important protein.
Although it is not yet possible to determine the dissociation constant for cryptophycin binding to tubulin, cryptophycin, and rhizoxin demonstrated similar potencies for reducing the binding of [ 3 H]vinblastine to tubulin (Fig. 8A). Since direct binding studies with [ 3 H]rhizoxin have suggested a k D of approximately 0.2 M and a stoichiometry of 1 for that drug (57), it is likely that the affinity of cryptophycin for this site is similar. We are currently synthesizing 3 H-labeled cryptophycin to directly determine the kinetic constants and stoichiometry of its association with tubulin. It is notable that cryptophycin is significantly more potent than vinblastine and rhizoxin with regard to its antimicrotubule effects in intact cells. For example, treatment of cells with only 25 nM cryptophycin blocked paracrystal formation in response to 5 M vinblastine, while 5 M rhizoxin was required for similar competition with vinblastine (Fig. 9). The roles of differences in rate constants for drug binding to and dissociation from tubulin and/or differential intracellular accumulation of the drugs are not yet known.
The apparent irreversibility of the effects of cryptophycin on microtubules both in intact cells and in vitro may prove useful for structural studies. Because of the propensity of high concentrations of tubulin to form amorphous aggregates, the three-dimensional structure of this protein has remained unsolved by either crystallographic or NMR technologies. Complexes of cryptophycin and tubulin may form crystals suitable for x-ray analysis.
In summary, the newly described antimitotic agent cryptophycin is able to interact with the Vinca alkaloid-binding site of tubulin with high potency and efficacy. This interaction causes markedly substoichiometric and virtually irreversible inhibi-tion of microtubule polymerization. Therefore, while the overall effects of cryptophycin are similar to those of other antimitotic drugs, the molecular details of its interaction with tubulin are unique.