Vol. 265, No. 28, Issue of October 5, pp. 17141-17149,199O Printed m U.S.A.

Binding of Dolastatin 10 to Tubulin at a Distinct Site for Peptide Antimitotic Agents Near the Exchangeable Nucleotide and Vinca Alkaloid Sites*

(Received for publication, March 26, 1990)

Ruoli Bai$, George R. Pettitj, and Ernest HamelSY From the $Laborutory of Molecular Pharmacology, Developmental Therapeutics Program, Diuision of Cancer Treatment, National Cancer Institute, National Institutes of Health, Bethesda, Marylend 20892 and the §Cancer Research Institute and Department of Chemistry, Arizona State University, Tempe, Arizona 85287

Dolastatin 10, a potent antimitotic peptide from a marine animal, strongly inhibits microtubule assem- bly, tubulin-dependent GTP hydrolysis, and the bind- ing of vinca alkaloids to tubulin. In studies of the binding of [3H]vincristine to the protein, with vinblas- tine as a control for competitive inhibition (Ki, 6.6 PM), we found that the macrolide antimitotic agents may- tansine and rhizoxin were also competitive inhibitors (Ki values, 3.1 and 12 PM). Dolastatin 10 and an un- related peptide antimitotic, phomopsin A, were more potent but noncompetitive inhibitors (Ki values, 1.4 and 2.8 PM). Since maytansine and, to a much lesser extent, vinblastine interfere with nucleotide exchange on tubulin, all drugs were examined for effects on nucleotide interactions at the exchangeable GTP site. Rhizoxin had effects intermediate between those of vinblastine and maytansine. Both peptides inhibited binding of radiolabeled GTP to tubulin even more strongly than did maytansine, but no drug displaced nucleotide from tubulin. The drugs were evaluated for stabilizing effects on the colchicine binding activity of tubulin. The peptides prevented loss of this activity, and vinblastine provided partial protection, while rhi- zoxin and maytansine did not stabilize tubulin. A tri- peptide segment of dolastatin 10 also effectively inhib- its tubulin polymerization and GTP hydrolysis. The tripeptide did not significantly inhibit either vincris- tine binding or nucleotide exchange, nor did it stabilize colchicine binding. These findings are rationalized in terms of a model with two distinct drug binding sites in close physical proximity to each other and to the exchangeable GTP site on &tubulin.

Pettit et al. (l-4) have described the isolation and charac- terization of several small peptides and depsipeptides, all containing unique amino acid residues, from the shell-less marine mollusk Dolabella auriculuria. The most potent of these molecules, dolastatin 10 (structure in Fig. l), inhibits mitosis in cells in culture and microtubule assembly in vitro (5). Structurally, dolastatin 10 consists of 4 amino acid resi-

* This work was supported in part by the Fannie E. Rippel Foun- dation, the Arizona Disease Control Research Commission, the Rob- ert B. Dalton Endowment Fund, and National Cancer Institute Outstanding Investigator Grant CA44344-OlAl (to G. R. P.). 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 U.S.C. Section 1734 solely to indicate this fact.

llTo whom correspondence should be addressed: Bldg. 37, Rm. 5C25, NIH, Bethesda, MD 20892.

dues (dolavaline, valine, dolaisoleuine, dolaproine) linked to a complex primary amine (dolaphenine). In these initial stud- ies (5) we also found that dolastatin 10 inhibited the binding of radiolabeled vinblastine and vincristine to tubulin and the tubulin-dependent hydrolysis of GTP. These observations led us to initiate a study comparing this new agent to vinblastine and to other drugs reported to inhibit the interaction of vinca alkaloids with tubulin, specifically maytansine (6), rhizoxin (7), and phomopsin A (8). The structures of these agents, as well as selected related compounds, are summarized in Fig. 2. We confirmed (5) that all agents inhibited tubulin polymeri- zation at substoichiometric drug concentrations (dolastatin lO/phomopsin A/vinblastine most active, rhizoxin least active on a molar basis) and that all agents inhibited the binding of radiolabeled vinblastine and vincristine to tubulin (dolastatin 10 most active, rhizoxin least active). We also demonstrated that these drugs inhibited tubulin-dependent GTP hydrolysis (dolastatin lO/phomopsin A/maytansine most active, rhi- zoxin least active).

It is not established whether these compounds interfere with the binding of the vinca alkaloids to tubulin by the same mechanism. Mandelbaum-Shavit et al. (6) found that vincris- tine and maytansine were competitive inhibitors of each oth- er’s binding to tubulin. Takahashi et al. (7) concluded that ansamitocin P-3, a close analog of maytansine, was a compet- itive inhibitor of the binding of radiolabeled rhizoxin to tu- bulin; but that rhizoxin and vinblastine were not competitive inhibitors of each other. No detailed kinetic studies of phom- opsin A inhibition of vinca alkaloid binding have been re- ported previously.

In addition to this group of drugs’ inhibitory effect on tubulin-dependent GTP hydrolysis, which could simply result from their inhibition of the polymerization reaction, there is increasing evidence of a close connection between the vinca domain’ of tubulin and the exchangeable GTP site localized to P-tubulin. Ludueiia and his collaborators (9-12) have dem- onstrated formation of an intra$-sulfhydryl cross-link be- tween Cys-12 and Cys-201 (or Cys-211) which occurs only in tubulin depleted of exchangeable nucleotide. Formation of this cross-link is weakly inhibited by vinblastine and strongly inhibited by maytansine and phomopsin A. More directly, Huang et al. (13) found that superstoichiometric amounts of

’ We have found, using [“Hlvincristine, that the drugs shown in Figs. 1 and 2 demonstrate two patterns of inhibition, competitive and noncompetitive, against the vinca alkaloids, implicating two distinct drug binding sites. Additional evidence to be described suggests these sites may be adjacent to each other. We therefore use the term “vinca domain” to include the “vinca site” (where the vinca alkaloids bind) together with regions of the tubulin molecule in close proximity to the site which strongly affect binding of vinca alkaloids.

17141

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

17142 Distinct‘Tubulin Site for Peptide Antimitotics

DOLASTATIN lo Oolawline Wine Dolaiedeuine Dol.vmins Oohdmnine

TRIPEPTIDE A

/ II

FIG. 1. Structural formulas of dolastatin 10 and its pro- tected synthetic precursor, tripeptide A. Note the nine asym- metric carbon atoms in dolastatin 10, five of which are also present in the same configurations in tripeptide A.

vinblastine weakly inhibited GDP and GTP exchange on tubulin, while maytansine almost totally inhibited nucleotide exchange. Neither drug, however, displaced exchangeable site nucleotide from tubulin.

In this report we describe studies which document that vinblastine, maytansine, and rhizoxin competitively inhibit the binding of radiolabeled vincristine to tubulin, while dolas- tatin 10 and phomopsin A are potent noncompetitive inhibi- tors. The two peptides equal or surpass maytansine in potency as inhibitors of nucleotide exchange, but they, too, do not displace bound nucleotide from tubulin. Rhizoxin affects nu- cleotide exchange only at high drug concentrations. We have also found that a tripeptide analog of dolastatin 10 (“tripep- tide A”; structure in Fig. 1) inhibits both microtubule assem- bly and GTP hydrolysis, but it has little effect on either nucleotide exchange or the binding of radiolabeled vincristine to tubulin. These observations are interpreted in terms of a model which envisages three closely related ligand binding sites largely confined to P-tubulin.

EXPERIMENTAL PROCEDURES

Materials-Maytansine (NSC 153858) and rhizoxin (NSC 332598) were obtained from the Natural Products Branch, National Cancer Institute. Phomopsin A was a generous gift of Dr. C. C. J. Culvenor (CSIRO Division of Animal Health, Parkville, Victoria, Australia) (14). Dolastatin 10 and tripeptide A were synthesized as described

previously (15). [“HlVincristine was a generous gift of Dr. D. G. Johns (6). Electrophoretically homogeneous bovine brain tubulin containing either nonradiolabeled GDP or [8-W]GDP in the exchangeable site and heat-treated microtubule-associated proteins were prepared as described elsewhere (16, 17). [8-‘%]GTP was obtained from Moravek Biochemicals, [3H]colchicine and [“Hlvinblastine from Amersham Corp., and nonradiolabeled vinblastine, colchicine, and GTP from Sigma. Nucleotides were repurified by anion exchange column chro- matography. Drugs were dissolved in dimethyl sulfoxide.

Methods-The binding of radiolabeled vincristine and nucleotides to tubulin was followed by the centrifugal gel filtration method described in detail previously (18). One ml columns of Sephadex G- 50 (Superfine) were used. The data from the vincristine binding experiments were analyzed by the standard methods described by Dixon et al. (19). The binding of radiolabeled colchicine to tubulin was followed on DEAE-cellulose filters (20), using the reaction con- ditions of Ludueiia et al. (21). Tubulin polymerization was followed turbidimetrically (22) at 350 nm in Gilford spectrophotometers

equipped with electronic temperature controllers. Base lines were established with the cuvette contents held at 0 “C, and the electronic temperature controller was set at 37 “C at zero time (the temperature in the cuvettes rose at about 0.5 “C/s). GTP hydrolysis was measured by following the generation of [8-‘%]GDP from [8-WJGTP by thin- layer chromatography on polyethyleneimine-cellulose as described in detail previously (18).

RESULTS

Inhibition of Vincristine Binding to Tub&in by Dolastatin 10, Phomopsin A, Vinblastine, Maytansine, and Rhizoxin- Previous studies had demonstrated that the antimitotic agents presented in Figs. 1 and 2 inhibited the binding of vinca alkaloids to tubulin (5-9). In an effort to determine whether one or more drug binding sites’ was involved in this inhibition, an extensive series of experiments was performed in which inhibition of the binding of [3H]vincristine was examined. The data obtained are summarized in Figs. 3 (dolastatin 10 and phomopsin A as inhibitors) and 4 (maytansine, rhizoxin, and vinblastine as inhibitors). Besides the standard Line- weaver-Burk presentation of the data (left panels), the results of these experiments are also presented in the Hanes format (rightpanels) (19). This second graphical method more readily distinguishes between noncompetitive and competitive inhi- bition since it generates curves which intersect on the abscissa in the former case and parallel curves in the latter case.

These experiments, and a second, confirmatory set (in the case of rhizoxin, three experiments were performed), demon- strated noncompetitive inhibition for the two natural product peptides and competitive inhibition for vinblastine (chosen as a positive control for competitive inhibition), maytansine, and rhizoxin. The data from these experiments were also analyzed by the Dixon method (19) to obtain Ki values, and these are summarized in Table I. The values obtained range from 1.4 pM for dolastatin 10, the most potent inhibitor, to 12 pM for rhizoxin, the least potent.

Drug Effects on the Binding of Guanosine Nucleotides at the Exchangeable Site-In a previous report (13) the effects of vinblastine and maytansine on the binding and displacement of nucleotide at the exchangeable site were described. Super- stoichiometric amounts of maytansine almost eliminated binding of radiolabeled GDP and GTP to tubulin-GDP, while vinblastine had substantially weaker effects. Using tubulin- [8-‘%]GDP and nonradiolabeled GDP or GTP we demon- strated that this phenomenon did not involve displacement of exchangeable site nucleotide from tubulin, but rather rep- resented inhibition of nucleotide exchange. Moreover, this inhibition of the exchange reaction was maximal at 0 “C and progressively declined as the reaction temperature was raised. Thus, maytansine binding to tubulin must be more rapid than nucleotide exchange, since maytansine binding, like nucleo- tide binding, is reversible (6). Equilibrium between all ligands was more closely approached at 37 “C as compared with 0 “C within the time frame of the experiments.

We wished to extend this analysis of drug effects on the exchangeable site to the newer vinca domain drugs dolastatin 10, phomopsin A, and rhizoxin, especially since these agents (5), like vinblastine (23) and maytansine (24), inhibit tubulin- dependent GTP hydrolysis. Table II presents data in which

’ Considerable disagreement exists in the literature regarding the number of vinca sites on tubulin and their relative affinities for the vinca alkaloids (see Ref. 32 for a review). The kinetic data on inhibition of vincristine binding we present here is readily interpreted in terms of a single vinca binding site. If more than one site exists, then the reaction conditions we have chosen probably favor binding almost exclusively at a single high affinity site to account for our uncomplicated results.

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Distinct Tub&n Site for Peptide Antimitotics

FIG. 2. Structural formulas of vinblastine, vincristine, vindesine, maytansine, ansamitocin P-3, an- samitocin P-4, rbizoxin, and phom- opsin A.

VINBLASTINE: Rl = CH3; R2= OCH3; R3= COCH3 VINCRISTINE: Rl= CHO; R2= OCH,; R3= COCH3

VINDESINE: Rl= CH3; R2= NHz; R3= H

MAYTANSINE: R= CH(CH3 INICH3 KOCH3 ANSAMITOCIN P-3: R= CHICH3h ANSAMITOCIN P-4: R= CHzCH(CH3)z

~)a CH3

OCH3

RHIZOXIN

H c’

CH3

iI \CH H, ,$O-;H-T\ 2

CH3NH---C IO 6CO I

H-C”

22

CO-N-&O-iH-C-:O-;H-;-CO,H

CH3 CHzCH3 H C02H 27

PHOMOPSIN A

the binding of 50 pM [&‘C]GTP to 10 pM tubulin-GDP was examined at 0 and 37 “C.

As before (13), a substoichiometric concentration of may- tansine did not affect the binding of the [&W]GTP to tubulin, but 5 pM phomopsin A reduced binding by almost 50% in the cold, indicating a rapid and tight binding of the drug to the protein. Dolastatin 10 at 5 pM had a much smaller effect on the binding of [8-W]GTP to tubulin at 0 “C. At a superstoichiometric drug concentration (20 pM) maytansine extensively inhibited the binding of [8-W]GTP to tubulin, as before (13), while the effect of vinblastine was again min- imal. At 20 pM rhizoxin has reproducibly had no effect. At 100 pM drug, however, rhizoxin has repeatedly surpassed vinblastine in the extent of inhibition of [8-Y]GTP binding.3

’ This may simply result from a weaker interaction of rhizoxin, as compared to the other drugs, with tubulin (e.g. K: values; also see Ref. 5). Relative to vinblastine, for example, equivalent inhibition of tub&n polymerization required a 4-fold higher concentration of

Identical experiments were also performed in which the binding of [8-“C]GTP to tubulin was examined at 37 “C (Table II). The inhibitory effects of maytansine, rhizoxin, and vinblastine on GTP binding disappeared at the higher tem- perature, while that of dolastatin 10 was substantially re- duced. Only phomopsin A retained its ability to inhibit GTP binding at 37 “C, with only a minimal decrease in the extent of inhibition observed with both the substoichiometric (5 pM)

and superstoichiometric (20 pM) concentrations examined. Since dolastatin 10 and phomopsin A must bind at a differ-

ent site(s) on tubulin than does maytansine, a conclusion based on the noncompetitive as opposed to competitive inhi- bition of vincristine binding, it is at least theoretically possible that these peptide antimitotics bind at the exchangeable nucleotide site. To examine this possibility the vinca domain drugs were also examined at 0 “C in an experimental system

rhizoxin; and equivalent inhibition of GTP hydrolysis required a 3- fold higher concentration (5).

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Distinct Tubulin Site for Peptide Antimitotics

10

& 5

2 4 D. 10

. .

/&!i!z 5 A

2 4

#,- VINCRISTINE (l/SI ,,M VINCAISTINE ISI

FIG. 3. Noncompetitive inhibition of the binding of [3H] vincristine to tubulin by dolastatin 10 (A and B) and by phomopsin A (C and D). Lineweaver-Burk analyses are presented in A and C and Hanes analyses in B and D. Each 0.6-ml reaction mixture contained 0.5 mg/ml tubulin, 0.1 M MES4 (pH 7.0 with NaOH in 1 M stock solution), 0.5 mM MgCl,, 5% (v/v) dimethyl sulfoxide, the indicated concentration of [“Hlvincristine, and inhibi- tors as summarized below. After a 20-min incubation at 22 “C, quad- ruplicate 0.15.ml aliquots of each reaction mixture were applied to syringe-columns and processed by centrifugal gel filtration as de- scribed previously (18) at room temperature. For the experiment presented in A and B the following concentrations of dolastatin 10 were used: A, none; 0, 1 gM; 0, 2 gM. For the experiment presented in C and D the following concentrations of phomopsin A were used: A, none; 0, 2 PM; 0, 3 FM. All lines were drawn by linear regression analysis of the data. Data points not shown in A and C (those obtained with 1 pM vincristine) were off-scale, but they were also used in the linear regression analysis. The conclusion that these data represent noncompetitive inhibition was confirmed by computer-assisted anal- yses of the data.

containing nonradiolabeled GTP and tubulin-[8-14C]GDP (Table III), but otherwise identical to the nucleotide exchange conditions described above. For the experiment summarized in Table III, it should be emphasized that a high amount of tubulin-bound nucleotide indicates the failure of nucleotide exchange and, moreover, retention of [8-14C]GDP in the pres- ence of drug. The data of Table III are complementary to those of Table II. Only where little [8-14C]GTP binding oc- curred (with 20 ~.LM dolastatin 10, phomopsin A, or maytan- sine) did little displacement of [8-14C]GDP occur. Conversely, under conditions where maximal binding of [8-14C]GTP oc- curred (no drug, 5 pM maytansine, or 20 pM rhizoxin) maximal displacement of [8-14C]GDP from tubulin occurred. This ex- periment allows us to conclude that, like maytansine, the peptide antimitotic agents inhibit nucleotide exchange with- out binding directly in the exchangeable site, since prebound GDP is retained by tubulin.

Properties of an Active Tripeptide Segment of Dolastatin IO-The total synthesis of dolastatin 10 (15), in part because of the peptide’s nine asymmetric carbon atoms, resulted in the preparation of a large number of isomers and synthetic intermediates, which have been examined for potential inter- actions with tubulin (25) (see “Discussion”). Only one inter- mediate was found to have significant activity,5 the carboxyl- protected tripeptide A (see Fig. l), which contains the three

4 The abbreviations used are: MES, 4-morpholineethanesulfonate; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetic acid.

“In contrast to dolastatin 10 (5), however, tripeptide A has no cytotoxic activity against L1210 murine leukemia cells and is thus probably not an antimitotic agent (25).

A I I /, , , 0.2 0.4 2 4

,A,- VINCRISTINE Il/Sl pM VINCRISTINE 6)

FIG. 4. Competitive inhibition of the binding of [‘Hlvincris- tine to tubulin by maytansine (A and B), by rhizoxin (C and D), and by vinblastine (E and F). Lineweaver-Burk analyses are presented in A, C, and E, Hanes analyses in B, D, and F. Reaction conditions were as described in the legend of Fig. 3. For the experi- ment presented in A and B the following concentrations of maytan- sine were used: A, none; 0, 4 pM; 0, 8 pM. For the experiment presented in C and D the following concentrations of rhizoxin were used: A, none; 0, 10 PM; 0, 20 pM. For the experiment presented in E and F the following concentrations of vinblastine were used: A, none; 0, 3 FM; 0, 5 pM. All lines were drawn by linear regression analysis of the data. Data points not shown in A, C, and E (primarily those obtained with 1 j&M vincristine) were off-scale, but they were also used in the linear regression analysis. The conclusion that these data represent competitive inhibition was confirmed by computer- assisted analyses of the data.

TABLE I

K, values for inhibitors of PHlvincristine binding to tub&n The data from Figs. 3 and 4, and from a second experiment with

all drugs except rhizoxin, with which three experiments were per- formed, were evaluated by the Dixon method (19) to obtain K, values. Average values are presented.

Drug Type of inhibition K,

PM Dolastatin 10 Noncompetitive 1.4” Phomopsin A Noncompetitive 2.8 (+0.3)b Maytansine Competitive 3.1 (&O.l)b Vinblastine Competitive 6.6 (+0.6)b Rhizoxin Competitive 12 (k3)’

a Identical values were obtained from the data of both experiments. bAverages of two values, with the deviations from the averages

presented in parentheses. ‘Average of three values, with standard deviation presented in

parentheses.

amino-terminal residues of dolastatin 10 intact. We have attempted to compare agents which inhibit tubulin

polymerization on a quantitative basis by determining the concentrations which result in a 50% reduction (IC& value) in the extent of the reaction. In our standard assay the reaction is followed for 20 min at 37 “C, and we examine the polymerization of purified tubulin dependent on 1 M gluta- mate and GTP (5, 26-28). The ICSo values obtained for the

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Distinct Tub&in Site for Peptide Antinitotics 17145

TABLE Ii Drug effects on the binding of [8-‘“C]GTP to tubulin-GDP

Each 0.5-m] reaction mixture contained 1.0 mg/ml (10 FM) tubulin- GDP, 0.1 M MES (pH 6.6 with NaOH in 1 M stock solution), 10% (v/v) dimethyl sulfoxide, 0.5 mM MgCL, and drug as indicated. Following addition of drug (inhibition of nucleotide binding requires that drug be added prior to the nucleotide (13)), 50 fiM [8-“CIGTP was added to each reaction mixture. Reaction mixtures were incu- bated at the indicated temperature for 15 min, 0.15-m] aliquots were applied to triplicate syringe-columns, and centrifugal gel filtration was performed as described previously (18) in a 4 “C cold room or 37 “C warm room, as appropriate. The experiments were performed at least three times, and average values, together with standard deviations (SD.), are presented.

Reaction temperature Drug added (+A)

0 “C 31 “C

pm01 [&“CIGTP boundlpmol tu- bulin W.D.!

None 0.72 (kO.03) 0.78 (kO.09) Dolastatin 10 (5) 0.46 (kO.16) 0.65 (f0.16) Phomopsin A (5) 0.35 (r0.06) 0.44 (kO.06) Maytansine (5) 0.69 (kO.07) 0.78 (kO.06) Dolastatin 10 (20) 0.13 (f0.12) 0.28 (~0.18) Phomopsin A (20) 0.05 (kO.03) 0.10 (+0.08) Maytansine (20) 0.09 (kO.05) 0.68 (kO.09) Vinblastine (20) 0.53 (kO.12) 0.77 (+0.05) Rhizoxin (20) 0.69 (kO.05) 0.77 (kO.09) Vinblastine (100) 0.54 (kO.05) 0.73 (kO.05) Rhizoxin (100) 0.16 (kO.09) 0.71 (t0.10)

TABLE III Drug effects on the displacement of [8-YJGDP from the

exchangeable site Each O.&ml reaction mixture contained the same components as

described for Table II except that 1.0 mg/ml (10 PM) tubulin-[8-14C] GDP was used. Following addition of the indicated drug (inhibition of nucleotide binding requires that drug be added prior to the nucleo- tide (l3)), 50 pM nonradiolabeled GTP was added to each reaction mixture. Reaction mixtures were incubated on ice for 15 min, and then processed as described in the legend of Table II.

Drug added pm01 [S-‘C]GDP (PM) bound/pmol tubulin

None (no nonradiolabeled GTP) 0.72 (kO.07)” None 0.03 (*o.ol)” Dolastatin 10 (5) 0.22 (?0.03)6 Phomopsin A (5) 0.35 (+0.06)* Maytansine (5) 0.04 Dolastatin 10 (20) 0.66 (kO.12) Phomopsin A (20) 0.72 (+-0.06)” Maytansine (20) 0.70 (+o.ll)” Vinblastine (20) 0.18 (+0.06)” Rhizoxin (20) 0.11 (to.11)” Vinblastine (100) 0.36 (+0.03)” Rhizoxin (100) 0.58 (kO.08)

” The averages from at least three experiments are presented, and the values in parentheses are the standard deviations.

’ The averages from two experiments are presented, and the values in parentheses are the deviations from the average.

’ Two experiments were performed with the same value obtained in both.

vinca domain drugs (5) were 1.2 PM for dolastatin 10, 1.4 FM for phomopsin A, 1.5 HM for vinblastine, 3.5 /.LM for maytan- sine, and 6.8 PM for rhizoxin. The analogous value for tripep- tide A was 4.2 PM. Fig. 5A presents representative data comparing inhibition of the glutamate-dependent polymeri- zation reaction by 2, 3, 4, and 5 WM tripeptide A to the inhibition obtained with 1 and 2 PM dolastatin 10. Tripeptide A was also compared to dolastatin 10 as an inhibitor of polymerization dependent on heat-treated microtubule-asso- ciated proteins (Fig. 5B), and relative activities of the two compounds were similar to those in glutamate.

0.3 0.4

;a 0.2

Q E Q 0.2 I

0.1

10 20 10 20 MINUTES

FIG. 5. Comparison of tripeptide A to dolastatin 10 as an inhibitor of the polymerization of purified tubulin in gluta- mate (A) and of microtubule assembly dependent on heat- treated microtubule-associated proteins (B). In the experiments presented in A, each 0.25-ml reaction mixture contained 1.0 mg/ml tubulin, 1.0 M monosodium glutamate (pH adjusted to pH 6.6 with HCI in a 2 M stock solution), 1 mM MgClz, 0.4 mM GTP, 4% (v/v) dimethyl sulfoxide, and drugs as follows: curue I, none; curue 2, 2 PM tripeptide A; curue 3, 3 PM tripeptide A; curue 4, 4 /IM tripeptide A; curve 5, 5 pM tripeptide A; curue 6, 1 pM dolastatin 10; curve 7, 2 FM dolastatin 10. All components except GTP were preincubated for 15 min at 37 “C and chilled on ice. The nucleotide was added, and polymerization was initiated by warming samples in the spectropho- tometer as described in the text. In the experiments presented in B, each 0.25-ml reaction mixture contained 1.5 mg/ml tubulin, 0.5 mg/ ml heat-treated microtubule-associated proteins, 0.1 M MES (ad- justed to pH 6.6 with NaOH in 1 M stock solution), 0.5 mM MgCL 0.5 mM GTP, 0.6% (v/v) dimethyl sulfoxide, and drugs as follows: curue 1, none; curve 2, 2 fiM tripeptide A; curve 3, 4 FM tripeptide A; curve 4, 6 PM tripeptide A; curue 5, 1 PM dolastatin 10; curue 6, 2 FM dolastatin 10. There was no preincubation with these samples. Poly- merization was initiated by warming the samples in the spectrophoto- meter as described in the text.

Tripeptide A, like dolastatin 10, strongly inhibited tubulin- dependent GTP hydrolysis (Table IV, Experiment A). When compared at 50 pM (concentrations which would totally in- hibit polymerization) both compounds inhibited formation of GDP by over 90%.

The effects of dolastatin 10 and tripeptide A on ligand binding, however, were totally different. High concentrations of tripeptide A had little or no effect on the binding to tubulin of [3H]vinblastine or [3H]vincristine (Table IV, Experiment B) or of [&‘“C]GTP (Table IV, Experiment C). Moreover, tripeptide A did not prevent displacement of [8-“C]GDP from tubulin by nonradiolabeled GTP (Table IV, Experiment D), as compared to the potent effect of dolastatin 10.

Stabilization of the Colchicine Binding Activity by Vinca Domain Drugs-As will be discussed in greater detail below (see “Discussion”), the above findings are consistent with dolastatin 10 and phomopsin A binding at the same site on tubulin (the “peptide site”). The evidence for this conclusion is completely indirect, based on their nearly identical prop- erties as inhibitors of vincristine binding and guanosine nu- cleotide exchange. Luduena et al. (21) have recently described an additional property of phomopsin A, near total stabiliza- tion of the colchicine binding activity of tubulin. If dolastatin 10 binds at the same site on tubulin as phomopsin A, it might share this property. We therefore compared the stabilizing effects of the two peptide natural products, and we also examined the other vinca domain drugs and tripeptide A at the same time (Table V).

Our results with phomopsin A and vinblastine are in excel- lent agreement with the data of Ludueiia et al. (21). Without a preincubation both phomopsin A and dolastatin 10, as well as vinblastine, moderately enhanced the binding of [3H]col- chicine to tubulin, while maytansine, rhizoxin, and tripeptide A had little effect on the reaction.

With a 3-h preincubation of tubulin at 37 “C prior to

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

17146 Distinct Tub&in Site for Peptide Antimitotics

TABLE IV Comparison of tripeptide A to dolastatin 10 in their effects on

interactions of tub&n with other ligands

Drug added Tubulin-dependent GTP hydrolvsis”

Experiment A None Tripeptide A Dolastatin 10

Drug added (PM)

nmol GDP formed

18.7 1.6 1.5

Inhibition of [“Hlvinblastine and [“H]vincristine bindin&

FHlVinblastine I”HlVincristine

Experiment B None Tripeptide A (50) Dolastatin 10 (2) Dolastatin 10 (5)

pm01 vinca alkaloid bound/pmol tub&in

0.17 0.12 (f0.01)’ 0.15 0.12 (kO.02)’

0.06 (+0.02)’ 0.03

Drug added bM)

Inhibition of [8-“C]GTP bindin&

Experiment C None Tripeptide A (100) Dolastatin 10 (20)

pm01 [8-“C]GTP bound/pmol tub&in

0.62 (r0.03)’ 0.67 (kO.04)’ 0.13 (+0.12)’

Drug added Inhibition of displacement of (FM) [8-W]GDP from tub&n’

pm01 [8-“C/GDP bound/pmol tub&in

Experiment D None (no nonradiolabeled GTP) 0.72 (+0.07)’ None 0.03 (&O.Ol)’ Tripeptide A (100) 0.07 (+0.05)’ Dolastatin 10 (20) 0.66 (+0.12)’

’ Each 50-~1 reaction mixture contained 1.0 mg/ml (10 PM) tubulin, 1.0 M monosodium glutamate (pH 6.6 with HCl in 2 M stock solution), 100 FM [8-i4C]GTP, 5% (v/v) dimethyl sulfoxide, and the indicated inhibitor at 50 FM. Incubation was at 37 “C for 30 min. An aliquot of each reaction mixture was processed for quantitation of the amount of GDP formed (expressed as nanomoles/ml of reaction mixture), as described previously (18).

b The experiment was performed as described in the legend of Fig. 3, with [“Hlvinblastine or [3H]vincristine at a concentration of 5 PM.

’ Average of at least three experiments, with standard deviation in parentheses.

d The experiments with tripeptide A were performed at the same time as those presented in Table II, and the rest of the data are from Table II.

e The experiments with tripeptide A were performed at the same time as those described in Table III, and the rest of the data are from Table III.

colchicine addition, in the absence of a second drug, subse- quent binding of [3H]colchicine was reduced by 44%. Vin- blastine provided good protection. In its presence 26% of the original binding activity was lost, while maytansine, rhizoxin, and tripeptide A provided no protection against decay of colchicine binding activity. As Ludueiia et al. (21) found, deterioration of colchicine binding activity did not occur in the presence of phomopsin A, and dolastatin 10 was equally effective in preserving this property of tubulin during a 37 “C incubation. We conclude that the results of this experiment support the concept of a distinct peptide site on tubulin, but the ineffectiveness of tripeptide A in stabilizing the colchicine binding activity complicates this explanation.

One possibility for the difference between the tripeptide and dolastatin 10 in stabilizing tubulin is a less extensive interaction of the tripeptide with the protein. This interpre-

TABLE V Effects of vinca domain drugs on the decay of the colchicine binding

activity of tub&n The reaction conditions used were those described by Luduena et

al. (21). Each O.l-ml reaction mixture contained 0.37 mg/ml tubulin, 0.1 M MES (pH 6.4 with NaOH from 1 M stock solution), 0.1 mM EDTA, 1 mM GTP, 0.5 mM MgC&, 1 mM 2-mercaptoethanol, 1 mM EGTA, 5% (v/v) dimethyl sulfoxide, and the indicated vinca domain drug at 50 PM. If indicated, the reaction mixtures were preincubated for 3 h at 37 “C and chilled on ice. [3H]Colchicine, 57 pM, was added to all reaction mixtures, and the samples were incubated for 2 h at 37 “C. The amount of colchicine bound to tubulin was determined by filtration through a stack of three DEAE-cellulose filters. Each re- action condition was performed in triplicate in each of three experi- ments. Average values with standard deviations (S.D.) in parentheses are presented.

Vinca domain drug added Not preincubated Preincubated

pm01 rH]cokhicine bound/pmol tub&in (S.D.)

None 0.34 (kO.04) 0.19 (kO.02) Dolastatin 10 0.47 (f0.03) 0.50 (kO.06) Tripeptide A 0.30 (f0.02) 0.16 (kO.03) Phomopsin A 0.52 (kO.06) 0.55 (f0.04) Vinblastine 0.43 (kO.05) 0.32 (kO.04) Maytansine 0.33 (kO.02) 0.16 (f0.02) Rhizoxin 0.32 (kO.02) 0.15 (fO.O1)

tation is supported by the tripeptide’s simultaneous loss of inhibitory effects on vincristine and GTP binding, but it is limited because of results obtained with the other three drugs. Only vinblastine had any stabilizing effect on tubulin. The other competitive inhibitors of vincristine binding, maytan- sine and rhizoxin, which also have stronger inhibitory effects than vinblastine on nucleotide exchange, had no stabilizing effect on colchicine binding activity. Thus, simple occupancy of the vinca site or interference with nucleotide exchange is insufficient to stabilize tubulin.

DISCUSSION

In this study we have examined effects on tubulin of vin- blastine and four additional antimitotic natural products of highly diverse structure. Two of these are peptides, dolastatin 10, derived originally from the marine animal D. auricularia (1) but now available by total synthesis (Xi), and phomopsin A, derived from the fungus Phomopsis leptostromiformis (14), and two are macrolides, maytansine, derived from several plants of the genus Maytenus (29) (the closely related ansa- mitocins are derived from a Nocarclia microorganism (30)), and rhizoxin, derived from the fungus Rhizopus chinensis (31).

These compounds had all been shown to inhibit the binding of vinca alkaloids to tubulin (5-8). Our initial goal was there- fore to determine whether all these agents were bound at a common site on tubulin. Previous studies had either not addressed this point or yielded seemingly contradictory re- sults. Mandelbaum-Shavit et al. (6) concluded that vincristine and maytansine were competitive inhibitors of each other’s binding to tubulin (both drugs were radiolabeled). Takahashi et al. (7), using radiolabeled rhizoxin, found that ansamitocin P-3 and vinblastine were competitive inhibitors of the binding of rhizoxin to tub&n; but they concluded that rhizoxin did not act as a pure competitive inhibitor of the binding of radiolabeled vinblastine to tubulin.

Our studies show that maytansine and rhizoxin, as well as vinblastine, are classic competitive inhibitors of the binding of radiolabeled vincristine to tubulin. We are thus in agree- ment with Mandelbaum-Shavit et al. (6) but not with Taka- hashi et al. (7) (these latter workers, however, evaluated rhizoxin inhibition of vinblastine binding). Our quantitative results differ from those of both groups. Our Ki value for

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Distinct Tub&in Site for Peptide Antimitotics 17147

maytansine as an inhibitor of vincristine binding is 3.1 pM as opposed to the value of 0.4 pM obtained by Mandelbaum- Shavit et al. (6). Takahashi et al. (7) found rhizoxin to be a potent inhibitor of vinblastine binding (K, value about 0.01 PM). We found it to be the weakest inhibitor in this group of drugs (K, value of 12 FM uersus vincristine), and we have also found similarly weak inhibition of vinblastine binding.

In contrast to the competitive inhibition of vincristine binding by maytansine and rhizoxin, the data obtained with the peptides dolastatin 10 and phomopsin A unambiguously demonstrated noncompetitive inhibition. In analyses of en- zymic catalysis a noncompetitive inhibitory pattern has gen- erally been interpreted to indicate binding of the inhibitor (i.e. dolastatin 10 or phomopsin A) at a different site from the substrate (i.e. vincristine) (19). Confidence in this conclu- sion with dolastatin 10 is greatly enhanced by our finding that tripeptide A fails to inhibit either vincristine or vinblas- tine binding to tubulin. The tripeptide probably binds in the portion of the dolastatin 10 site responsible for inhibition of tubulin polymerization in a manner that does not obstruct vinca alkaloid binding.

We believe dolastatin 10 and phomopsin A bind in the same site on tubulin, the peptide site. Structural analogies do exist between the two peptides, especially in hydrophobic amino acids that appear to be derived from valine and isoleucine, if the residues present in tripeptide A are the key features of dolastatin 10 (see below). More important, however, are the functional similarities of the two peptides. Both inhibit tu- bulin polymerization, tubulin-dependent GTP hydrolysis, and nucleotide exchange on tubulin, and both strongly stabilize the colchicine binding activity of tubulin. Definitive proof of this conclusion of a common binding site will require a direct method for measuring binding of at least one of these agents.

In this analysis the question of the number of vinca alkaloid binding sites on tubulin (reviewed in Ref. 32) has not been considered. If more than one site of different affinities is involved in the binding of these drugs, then it is probable that the inhibitory studies presented in Figs. 3 and 4 involved only a single high affinity site to account for the straight-forward results obtained. Although dolastatin 10, phomopsin A, rhi- zoxin, and maytansine could have similar quantitative effects on multiple vinca sites, this seems unlikely. Presumably the drug effects studied here involve a single high affinity vinca site.6

Dolastatin 10 and phomopsin A resemble maytansine in potently inhibiting nucleotide exchange on tubulin without displacing nucleotide present in the exchangeable site (13). Most notably, this property of dolastatin 10 is lost by its tripeptide precursor. As with maytansine, superstoichiometric concentrations of the antimitotic peptides are required for maximal inhibition of nucleotide exchange. Since polymeri- zation is completely inhibited at substoichiometric drug con- centrations at which effects on nucleotide binding are barely detectable, this effect cannot be the primary cause for these agents’ inhibition of microtubule assembly. The difference

’ In a previous study (38) one of us (E. H.) observed, by Scatchard analysis, evidence for two classes of binding site for [“Hlvinblastine, although the intercepts for both sites were substoichiometric. A similar result was obtained for [“Hlvincristine binding in the present set of experiments when Scatchard analysis was performed on data obtained over a wide range of vincristine concentrations. The studies presented in Figs. 3 and 4 were performed at relatively low vincristine concentrations, and the maximum stoichiometry observed in these studies in the uninhibited reaction mixtures was about 0.1 mol of vincristine bound per mol of tubulin. This further supports our conclusion that the effects reported here involve a single high affinity site.

CY

Exchangeable GTP SW

FIG. 6. Schematic model with proposed interrelationships on fl-tubulin of the vinca/maytansine/rhizoxin site, the peptide site binding dolastatin 10 and phomopsin A, and the ex- changeable GTP site. The peptide site is presented as a groove in which configurationally important chiral centers of dolastatin 10 are bound, while the dolaphenine (and perhaps dolaproine, which would require presenting the sites in a different orientation) moiety blocks access to the vinca and exchangeable nucleotide sites. The projection of dolastatin 10 presented here, with the side chains of the 3 amino- terminal residues and the carboxyl-terminal portion of the molecule (positions l-10) above the plane of the paper, was drawn from a molecular model of the peptide. The model also proposes that may- tansine, but not vinblastine, binds in the vinca site in a manner which physically blocks access to the exchangeable site by nucleotides.

between dolastatin 10 and tripeptide A confirms this conclu- sion.

There is one major difference between the effects of may- tansine and of the antimitotic peptides, particularly phom- opsin A, on nucleotide exchange. The effect of maytansine is dramatic at 0 “C but negligible at 37 “C. With phomopsin A especially, and with dolastatin 10, significant inhibition of nucleotide exchange does occur at 37 “C. The temperature difference with maytansine may be caused by the reversibility (6) of its binding to tubulin: rapid drug dissociation reactions at the higher temperature permit nucleotide exchange during the brief intervals when tubulin is free of maytansine. If this interpretation is correct, it predicts relatively slow dissocia- tion reactions for dolastatin 10 and especially phomopsin A. The latter, in fact, may bind to tubulin essentially irreversibly, for there is little difference in its inhibitory effects on nucleo- tide exchange as a function of temperature.

The inhibitory effects of vinblastine and rhizoxin on nu- cleotide exchange are much less dramatic than those of the other agents at the two concentrations examined. At 20 pM, rhizoxin had little effect on nucleotide exchange, while vin- blastine was slightly inhibitory. At 100 pM, however, the inhibition of nucleotide exchange observed with rhizoxin was much greater than that which occurred with vinblastine. As with maytansine, substantial inhibition of exchange by rhi- zoxin and vinblastine only occurred in the cold. Although a more extensive concentration study is required, we believe that the high rhizoxin requirement for inhibition of GTP binding reflects its weaker interaction with tubulin (mani- fested by its high K, value as an inhibitor of vincristine binding, as shown here, and its relatively high I& values as an inhibitor of polymerization and GTP hydrolysis (5)). Rhi- zoxin thus may be most similar to maytansine as an inhibitor of nucleotide exchange if its lower affinity for tubulin is considered.

As with polymerization, inhibition of GTP hydrolysis by

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

17148 Distinct Tub&in Site for Peptide Antimitotics

these agents cannot simply be ascribed to their effects on nucleotide exchange. Hydrolysis requires a higher reaction temperature, and, with the exception of phomopsin A, this eliminates or reduces inhibition of exogenous GTP binding at the exchangeable site. Moreover, with the possible excep- tion of rhizoxin, inhibition of hydrolysis is more extensive than of nucleotide exchange at the same drug concentration (cf. Ref. 5), as is the case with polymerization. Most notable, again, is the contrast between the effect of dolastatin 10 and its tripeptide analog. The latter binds to tubulin (presumably in the peptide site), inhibiting both polymerization and GTP hydrolysis, but, in contrast to dolastatin 10, it has minimal effect on nucleotide exchange.

More specific modeling of the peptide site is possible based on structure-activity studies we have performed with dolas- tatin 10 (25). The peptide has nine chiral centers (see Fig. l), and their absolute configurations were established by total synthesis of the peptide (15). Isomers of dolastatin 10 and of tripeptide A were also prepared. These had altered configu- rations at one or more of the following positions: 9, 10, 18, 19, and 19a (25). If carbons 9, 10, and/or 19a were in configura- tions opposite to those in dolastatin 10, inhibition of tubulin polymerization occurred which was comparable to that ob- served with dolastatin 10. Reversal of configuration at carbons 18 and/or 19, however, yielded weakly inhibitory peptides, while reversal of configuration at positions 18 and/or 19 in combination with reversal at positions 9 and/or 10 yielded noninhibitory peptides. Analogs of tripeptide A with reversal of configuration at positions analogous to carbons 18 and 19 were also inactive as inhibitors of tubulin polymerization.

Although allosteric factors cannot be excluded for the ligand interactions described here, we think it more likely that direct interference of the ligands with each other occurs due to physical proximity of their binding sites on the tubulin mol- ecule (see Fig. 6). The drug binding sites are probably pri- marily on @-tubulin, since the exchangeable nucleotide site has been localized to this subunit (33-36). The simplest model requires that the vinca site be adjacent to the exchangeable nucleotide site.’ When maytansine, and perhaps rhizoxin, binds in the vinca site, a portion of the drug molecule may physically overlie the exchangeable nucleotide and thereby impede nucleotide exit and/or entry. Such a model is also consistent with the findings of Ludueiia and his colleagues (g-12), who found that formation of a cross-link between Cys- 12 and Cys-201 (or -211) in /?-tubulin is strongly inhibited by guanine nucleotide, maytansine, or phomopsin A but not by vinblastine.

We further propose that the peptide site is distinct from, but in close proximity to, both the vinca site and the ex- changeable nucleotide site (Fig. 6). Studies with dolastatin 10 chiral analogs and amino-terminal tripeptide analogs (25) have demonstrated that proper configuration at positions 18 and 19, but not 19a, in the dolaisoleuine moiety are critical for maximum activity, and that configurations at positions 9 and 10 in the dolaproine moiety are relatively unimportant. The backbone of the first three residues of dolastatin 10 may lie in a hydrophobic pocket or groove with little or no overlap with either the vinca or exchangeable nucleotide site (since tripeptide A has limited effects on the binding of these li- gands). Therefore, the dolaproine and/or dolaphenine resi- dues could offer a physical barrier both to the binding of vinca alkaloids and to nucleotide exchange. Such a model predicts that dolastatin 10 is an extended molecule, but its crystal structure is unknown. A crystal structure for dolastatin 10

’ A photoaffinity analog of vinblastine, however, was specifically cross-linked to both a- and P-tubulin in a 3 to 2 ratio (38).

should also clarify its structural homologies with phomopsin A, for which such data are available (37).

The side chain of phomopsin A, known to project from the 13-member ring (37), may bind in the proposed peptide groove, with some component(s) of the ring obstructing access to the nucleotide and vinca sites, or vice versa. While struc- tural analogies do exist between dolastatin 10 and the side chain of phomopsin A, Lacey et al. (8) found that two analogs of phomopsin A with major modifications in the side chain were as active as phomopsin A as inhibitors of vinblastine binding and microtubule assembly: phomopsinamine A, which lacks the substituent at position 24, and octahydrophomopsin A, in which the four nonaromatic carbon-carbon double bonds are reduced. This reduction introduces three new chiral cen- ters in the side chain, implying that octahydrophomopsin A is a mixture of multiple, equally active isomers. Since proper configuration at positions 18 and 19 in dolastatin 10 are needed for maximal interaction of the drug with tubulin (25), it seems unlikely the protein would tolerate major modifica- tions in key structural features of phomopsin A. Clarification of the important structural analogies in these drugs will require additional structure-activity evaluations.

Besides the effects of vinblastine, maytansine, and phom- opsin A on /?-tubulin cysteine cross-links described above (9- 12), most other studies in the literature implicate the p- subunit as the binding site for vinca domain drugs. A photo- affinity analog of vinblastine formed specific (ie. inhibited by vinblastine) covalent bond(s) with /3-tubulin, although it in- teracted with the (Y subunit, too (38). A human rhabdo- myosarcoma cell line resistant to vincristine produces an altered p-tubulin (39), as does an Aspergillus nidulans mutant resistant to ansamitocin P-3 and rhizoxin (40). Implication of the P-subunit is not universal, however, for a Chinese hamster ovary cell line resistant to maytansine produces an altered a-tubulin (41).

Acknowledgments-We wish to thank Drs. D. L. Herald and S. B. Singh for assistance in this work.

REFERENCES

1. Pettit, G. R., Kamano, Y., Herald, C. L., Tuinman, A. A., Boett- ner, F. E., Kizu, H., Schmidt, J. M., Baczynskyj, L., Tomer, K. B., and Bontems, R. J. (1987) J. Am. Chem. Sot. 109, 6883- 6885

2. Pettit, G. R., Kamano, Y., Kizu, H., Dufresne, C., Herald, C. L., Bontems, R. J., Schmidt, J. M., Boettner, F. E., and Nieman, R. A. (1989) Heterocycles 28, 553-558

3. Pettit, G. R., Kamano, Y., Dufresne, C., Cerny, R. L., Herald, C. L.. and Schmidt. J. M. (1989) J. Orz. Chem. 64.6005-6006

4. Pet&, G. R., Kamano, Y.; Herald, C.-L., Dufresnk, C., Bates, R. B., Schmidt, J. M., Cerny, R. L., and Kizu, H. (1990) J. Org. Chem. 55,2991-2992

5. Bai, R., Pettit, G. R., and Hamel, E. (1990) Biochem. Phurmacol. 39,1941-1949

6. Mandelbaum-Shavit, F., Wolpert-DeFilippes, M. K., and Johns, D. G. (1976) Biochem. Biophys. Res. Commun. 72,47-54

7. Takahashi, M., Iwasaki, S., Kobavashi, H., Okuda, S., Murai, T., and Sato, Y..(1987) Biochim. Bjophys. Acta 926, 215-223

8. Lacey, E., Edgar, J. A., and Culvenor, C. C. J. (1987) Biockem. Pharmncol. 36,2133-2138

9. Roach. M. C.. and Ludueiia. R. F. (1984) J. Biol. Ckem. 269. 12063-12071

10. Luduena, R. F., and Roach, M. C. (1981) Arch. Biochem. Biophys. 210,498-504

11. Little, M., and Ludueiia, R. F. (1987) Biochim. Biopkys. Acta 912,28-33

12. Luduena, R. F., Roach, M. C., Prasad, V., and Lacey, E. (1990) Biochem. Pharmucol. 39,1603-1608

13. Huang, A. B., Lin, C. M., and Hamel, E. (1985) Biockem. Biophys. Res. Commun. 128,1239-1246

14. Culvenor, C. C. J., Beck, A. B., Clarke, A., Cockrum, P. A., Edgar,

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

Distinct Tubulin Site for Peptide Antimitotics 17149

15.

16. 17.

18.

19.

20. 21.

22.

23.

24. 25.

26.

27.

J. A., Frahn, J. L., Jago, M. V., Lanigan, G. W., Payne, A. L., Peterson, J. E., Petterson, D. S., Smith, L. W., and White, R. R. (1977) Aust. J. Biol. Sci. 30, 269-278

Pettit, G. R., Singh, S. B., Hogan, F., Lloyd-Williams, P., Herald, D. L., Burkett, D. D., and Clewlow, P. J. (1989) J. Am. Chem. Sot. 111,5463-5465

Hamel, E., and Lin, C. M. (1984) Biochemistry 23,4173-4184 Duanmu, C., Lin, C. M., and Hamel., E. (1986) Biochim. Biophys.

Acta 881, 113-123 Hamel, E., and Lin, C. M. (1984) J. Biol. Chem. 259, 11060-

11069 Dixon, M., Webb, E. C., Thorne, C. J. R., and Tipton, K. F.

(1979) Enzymes, pp. 332-381, Academic Press, New York Borisy, G. G. (1972) Anal. Biochem. 13,373-385 Luduena, R. F., Prasad, V., Roach, M. C., and Lacey, E. (1989)

Arch. Biochem. Biophys. 272,32-38 Gaskin, F., Cantor, C. R., and Shelanski, M. L. (1974) J. Mol.

Biol. 89,737-758 David-Pfeuty, T., Simon, C., and Pantaloni, D. (1979) J. Biol.

&em. 254,11696-11702 Lin, C. M., and Hamel, E. (1981) J. Biol. Chem. 256,9242-9245 Bai, R., Pettit, G. R., and Hamel, E. (1990) Biochem. Pharmucol.,

in press Batra, J. K., Kang, G.-J., Jurd, L., and Hamel, E. (1988) Biochem.

Pharmacol. 37,2595-2602 Lin, C. M., Singh, S. B., Chu, P. S. Dempcy, R. O., Schmidt, J.

M., Pettit, G. R., and Hamel, E. (1988) Mol. Pharmacol. 34, 200-208

28. Muzaffar. A.. Brossi. A.. Lin. C. M.. and Hamel. E. (1990) J.

29. Med. &em. 33,567-571 ’ ’

Kupchan, S. M., Komodo, Y., Court, W. A., Thomas, G. J., Smith, R. M., Karim, A., Gilmore, C. J., Haltiwanger, R. C., and Bryan, R. F. (1972) J. Am. Chem. Sot. 94, 1354-1356

Higashide, E., Asai, M., Ootsu, K., Tanida, S., Kozai, Y., Hase- gawa, T., Kishi, T., Sugino, Y., and Yoneda, M. (1977) Nature 270, 721-722

30.

31. Iwasaki, S.. Namikoshi, M., Kobavashi, H., Furukawa. J., Okuda,

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

S., It& A., Kasuya, A.; Iitaka, Y.; and Sato, Z: (1986) J. Antibiotics 39, 424-429

Hamel. E. (1990) in Microtubule Proteins (Avila. J.. ed) DD. 89- 191, CRC Press, Boca Raton, FL ~ ’ ’ --

Geahlen, R. L., and Haley, B. E. (1977) Proc. Natl. Acad. Sci. U. S. A. 74,4375-4377

Nath, J. P.. Eagle. G. R., and Himes. R. H. (1985) Biochemistry 24; 155511560

Hesse, J., Thierauf, M., and Ponstingl, H. (1987) J. Biol. Chem. 262,15472-15475

Linse, K., and Mandelkow, E.-M. (1988) J. Biol. Chem. 263, 15205-15210

Mackay, M. F., Van Donkelaar, A., and Culvenor, C. C. J. (1986) J. Chem. Sot. Chem. Commun. 1219-1221

Safa, A. R., Hamel, E., and Felsted, R. L. (1987) Biochemistry 26,97-102

Houghton, J. A., Houghton, P. J., Hazelton, B. J., and Douglass, E. C. (1985) Cancer Res. 45, 2706-2712

Takahashi, M., Kobayashi, H., and Iwasaki, S. (1989) Mol. & Gen. Genet. 220,53-59

Schibler, M. J., and Cabral, F. R. (1985) Can. J. Biochem. Cell Biol. 63,503-510

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from

R L Bai, G R Pettit and E Hamelnear the exchangeable nucleotide and vinca alkaloid sites.

Binding of dolastatin 10 to tubulin at a distinct site for peptide antimitotic agents

1990, 265:17141-17149.J. Biol. Chem. 

  http://www.jbc.org/content/265/28/17141Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/265/28/17141.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on March 13, 2018

http://ww

w.jbc.org/

Dow

nloaded from