Characterization of the Interaction of Cryptophycin 1

(CANCER RESEARCH 56. 4398-4406. October 1. 1996] Characterization of the Interaction of Cryptophycin 1 with Tubulin: Binding in the Vinca Domain, Competitive Inhibition of Dolastatin 10 Binding, and an Unusual Aggregation Reaction Ruoli Bai, Robert E. Schwartz, John A. Kepler, George R. Pettit, and Ernest Hamel1

Laboratory of Molecular Plitinmict)lt>xy. Division of Basic Sciences, National Cancer Institute, NIH, Bethcsila, Marciatiti 20XV2 Â¡K.H., E. H.l; Merck, Sharp unti Dohnie Research Laboratories. Railway, New Jersey 07065 [R. E. S.Â¡; Research Triangle Institute. Research Triangle Park, North Carolina 27709 Â¡J.A. K.]; ami Cancer Research Institute ami Department of Chemistry, Arizona State University, Tempe, ArÃ¬-onaR52H7 [G. R. P.I

ABSTRACT activity against a number of solid tumors implanted in mice, and CP1 or a closely related agent will probably be evaluated as an experi The Â¡miÂ¡miloticdepsipeptide cryptophycin l (CP1) was compared to mental antineoplastic agent in clinical trials. the antimitotic peptide dolastatin 10 (DIO) as an antiproliferative agent We have been studying interactions of antimitotic peptides (DIO and in its interactions with purified tubulin. The potent activity of CP1 as and analogues; Refs. 9-14) and depsipeptides (DIS and analogues; an inhibitor of cell growth was confirmed. The agent had an IC50 of 20 p.vt against L1210 murine leukemia cells versus 0.5 n.Mfor DIO. Both drugs Ref. \5f with purified tubulin, as well as the comparative antiprolif were comparable as inhibitors of the glutamate-induced assembly of erative activities of these agents (structures in Fig. 1). DIO binds in a purified tubulin, with DIO being slightly more potent. CP1, like DIO, was distinct region of the tubulin a-ÃŸ-heterodimer, since it noncompeti- a noncompetitive inhibitor of the binding of [3H]vinblastine to tubulin tively inhibits the binding of radiolabeled Vinca alkaloids to tubulin, (apparent Kâ€ž3.9/IMI; and the depsipeptide was a competitive inhibitor of as do an unrelated group of sponge-derived macrocyclic polyethers the binding of |'H]D10 to tubulin (apparent K,, 2.1 JUM).CP1 was less termed halichondrins (16) and spongistatins (17, 18). The binding of potent than DIO as an inhibitor of nucleotide exchange on tubulin, but the |'H]D10 to tubulin was inhibited competitively by one of its ana two drugs were equivalent in stabilizing the colchicine binding activity of logues [(19a/?)-isodolastatin 10] and by another peptide antimitotic tubulin. CP1, like DIO, caused the formation of extensive structured agent, phomopsin A, and noncompetitively by spongistatin 1 (18). aggregates of tubulin when present in stoichiometric amounts relative to With [3H]D10, we observed tenacious, but reversible, binding to the protein. Whereas at lower concentrations the drugs were equivalent in causing formation of small oligomers detected by gel permeation high- tubulin, and formation of drug-containing tubulin oligomers and performance liquid chromatography, there were notable differences in the coiled polymers that could be studied by HPLC, electron microscopy, aggregation reactions induced by the two drugs. The electron micro- or turbidity development (13, 18). DIO, moreover, potently inhibited graphic appearance of the DIO-induced aggregate differed substantially nucleotide exchange on tubulin, although it did not displace nucleo from that of the CPl-induced aggregate. With DIO, but not CP1, aggre tide bound in the exchangeable site. The peptide prevented time- gate morphology was greatly altered in the presence of microtubule- dependent decay of CLC binding activity of tubulin (10). associated proteins. Finally, although CP1 caused the formation of mas In contrast, the depsipeptide D15. despite its partial structural sive aggregates, as did DIO, there was little turbidity change with the analogy to DIO and its ability to inhibit tubulin polymerization, depsipeptide as opposed to the peptide. neither induces tubulin oligomerization nor affects the interactions of other ligands with tubulin (13, 15). Since we have not succeeded in INTRODUCTION demonstrating an interaction of |3H)D15 with tubulin,4 we have attributed this difference in properties to much weaker binding of D15 Antimitotic natural products have a long history of use as poisons and as therapeutic agents. Ever since CLC2 was used to isolate tubulin to tubulin. Our work with the dolastatins made us particularly curious about (1), the target protein for this class of drug, there has been a continu CP1. We wondered how its properties would compare with those of ing discovery of new agents remarkable for their chemical diversity the dolastatins. particularly whether its being a depsipeptide indi and often for their potent cytotoxicity (2). Smith el al. (3) reported that the depsipeptide CP1 (previously known as "cryptophycin" and cated a relatively weak interaction with tubulin. We found, how "cryptophycin A"; structure in Fig. 1, as reported by Barrow et al. in ever, that CP1 resembled DIO more than D15 in its effects on tubulin. The agent noncompetitively inhibited [ 'H|VLB binding to Ref. 4). a compound initially isolated as a potent antifungal agent tubulin and competitively inhibited |'H]D1() binding to tubulin. from the terrestrial cyanobacterium Nostoc sp. (5), caused cells to Like DIO, CPI caused formation of structurally distinct tubulin accumulate in mitosis with the disappearance of intracellular micro- aggregates, but this occurred without a change in the turbidity of tubules. IC50s as low as 2 pM were obtained for cell growth, and these the tubulin solution. CPI induced-aggregates differed morpholog values for CP1 were approximately 100-1000-fold lower than those ically from those induced by VLB and DIO under the same reaction obtained with paclitaxel (Taxol), CLC, and VLB. Moreover, CP1 was conditions. In addition, we confirmed the potent antimitotic activ a poor substrate for the multidrug resistance transport protein. ity reported by Smith et al. (3). Kerksiek el al. (6) and Smith and Zhang (7) subsequently found that CP1 inhibited tubulin assembly and the binding of VLB but not CLC to tubulin. Golakoti et al. (8) found that CP1 had excellent In vivo MATERIALS AND METHODS

Materials. CPI was purified from the culture medium used to grow the Received 4/8/96; accepted 7/30/96. Nostoc sp. (5). Synthetic DIO (19). |'H]D10 ( 13), synthetic D15 (20). ( 19aA)- The costs of publication of this article were defrayed in part by Ihe payment of page charges. This article must therefore be hereby marked advertisement in accordance with isodolastatin 10 (21), an isomer of DIO with reversal of configuration at 18 U.S.C. Section 1734 solely to indicate this fact. position C-19a, and electrophorelically homogeneous bovine brain tubulin and 1To whom requests for reprints should be addressed, at Laboratory of Molecular heat-treated MAPs (22) were prepared as described previously. [8-uC|GTP. Pharmacology. Division of Basic Sciences. National Cancer Institute. NIH, Building 37. Room 5C25, Bcthesda. MD 20892. Phone: <301) 496-4855; Fax: (301) 496-5839. - The abbreviations used are; CLC. colchicine; VLB. vinblastine; CP1. cryptophycin 1; 'G. R. Petlit, E. J. Flahive, M. R. Boyd. R. Bai. E. Hamel. and J. M. Schmidt, DIO. dolastatin 10; D15. dolastatin 15; HPLC. high-performance liquid chromatography: submitted for publication. MAPs. microlubule-associated proteins; MES. 4-morpholineethanesulfonate. 4 R. Bai. J. T. Mullaney. G. R. Pettil. and t. Hamel. unpublished observations. 439S

OCH,

CRYPTOPHYCIN 1

Fig. 1. Structural formulas fur CPI. DIO. and D15. 11 I m Ã¬\ S ^^ I O CH OCH O / X â€”¿N

DOLASTATIN 10

H3C CH3H3CÂ«Â«Â« CH3H3C CH3

CH,

DOLASTATIN 15

|'H|VLB. and |'H|CLC were obtained from Moruvek Biochemicals. Amer- Tubulin aggregation5 was followed by HPLC on a 7.5 x 300-mm TSK shum. and DuPont. respectively. Maytansine and rhizoxin were obtained from G3000SW gel permeation column with an LKB system in line with a Ramona the Drug Synthesis and Chemistry Branch, National Cancer Institute; CLC and 5-LS flow detector. The column was equilibrated with a solution containing VLB were from Sigma Chemical Co. All drugs were dissolved in DMSO. and 0.1 M MES (pH 6.9) and 0.5 IHMMgCK. Absorbance data were evaluated with equivalent amounts of solvent were added to control reaction mixtures. MES Raytest software on an IBM-compatible computer. Tubulin aggregation was was prepared as 1.0 Mstock solutions, adjusted to the indicated pH values with also followed turbidimetrically at 350 nm in Gilford spectrophotometers. NaOH: monosodium glutamate was prepared as a 2.0 M stock solution, Negatively stained specimens were examined in a Zeiss model IOCA adjusted to pH 6.6 with HCI. electron microscope. About 5 Â¿tlwere placed on a 200-mesh. carbon-coated. Methods. Inhibition of glutamate-induced assembly of purified tubulin Formavar-treated copper grid, and after 5-10 s. were washed off with 5-10 was evaluated in 0.25-ml reaction mixtures following a drug-tubulin pre- drops of 0.5% uranyl acetate. Excess stain was removed by absorbance into incubation for 15 min at 37Â°Cwithout GTP in 0.24 ml. Final concentrations torn filter paper. were 1.0 mg/ml (10 \M) tubulin, 1.0 M monosodium glutamate, 1.0 niM Drug effects on the growth of L1210 murine leukemia cells were deter MgCK. 0.4 mM GTP, and 47c (v/v) DMSO. Assembly was initiated by a mined on cells grown under 5% CO, in 5 ml of RPMI 1640 supplemented with 75-s jump from 0 to 37Â°Cand was monitored in Gilford spectrophotometers 17% fetal bovine serum. 2 HIMglutamine. and 10 /j.g/ml gentamicin sulfate. at 350 nm. The extent ot the reaction was evaluated after 20 min. In studies Cell number was determined with a Coulter Counter following a 24-h incu nl inhibition of assembly with MAPs, there was no drug-tubulin preincubation at 37Â°C.Mitotic index was evaluated by microscopic examination of bation. Reaction mixtures contained 1.5 mg/ml ( 15 /J.M)tubulin, 0.6 mg/ml cells fixed with methanol and stained with Giemsa. heat-treated MAPs. 0.5 mM GTP. 0.5 mM MgCl2, 4% DMSO, and 0.1 M MES (pH 6.4). The binding of (8-'4C]GTP. |'H|VLB, and |'H]D10 to tubulin were eval RESULTS uated by centrifugal gel filtration on microcolumns of Sephadex G-50 (super CPI Is a Potent Inhibitor of Cell Growth fine) prepared in tuberculin syringes (13). Since the GTP binding assay actually measures inhibition of nucleotide exchange (23). the radiolabeled GTP We have repeatedly obtained IC5()sof 0.3-0.9 DMfor DIO and 3-4 was the last component added to the reaction mixtures in the GTP binding nM for D15 as inhibitors of the growth of LI2IO murine leukemia assays. Stabilization of the CLC binding activity of tubulin was evaluated using the ' We are using the term "aggregation" to indicate tubulin polymeri/.alion into struc reaction conditions described by LudueÃ±aet al. (24). Reaction mixtures were tures with discrete morphology that do not contain linear protofilaments {i.e.. polymers filtered through a stack of three DEAE-cellulose filters. other than microtuhules and sheets or ribbons resembling open microtuhulesi. 4399

Effects of CPI on Interactions of Other Ligands with Tubulin

0.6 VLB and DIO. We, like Kerksiek et al. (6) and Smith and Zhang (7), found that CPI inhibited the binding of |'H]VLB to tubulin. In addition. CPI inhibited the binding of |'H]D1() to tubulin. Compari son with other drugs (Table 2), at a low inhibitor concentration (inhibitonradiolabeled ligand ratio, 1:2). showed that CPI was among 0.4 the stronger inhibitors of both reactions. Extensive kinetic analyses of inhibition of the radiolabeled drug binding reactions demonstrated a significant difference in the inhibi tory effects of CPI in the two reactions. Data are presented in the Hanes format for inhibition of pH]VLB binding (Fig. 3A) and inhi 0.2 bition of ['HIDIO binding (Fig. 3ÃŸ).With the Hanes format, inhibi tory data obtained at different inhibitor concentrations yield parallel lines for a competitive inhibitor and lines that intercept on the nega tive abscissa for a noncompetitive inhibitor (25). The data of Fig. 3 thus indicate that CPI is a noncompetitive inhibitor of |'H]VLB 20 20 binding to tubulin and a competitive inhibitor of |'H]D10 binding to MINUTES tubulin. By the terminology proposed earlier, CPI apparently binds in Fig. 2. Inhibition of tubulin assembly by substoichiometric concentrations of CP1. the "peptide site" of the "Vinca domain." Dixon analysis (25) yielded Reaction components were described in the text. For the experiments shown in both panels, at time /.ero the temperature was changed from 0Â°Cto 37Â°C.A, MAP-dependent apparent Kp for CPI of 3.9 JU.Mversus VLB and 2.1 /J.Mversus DIO assembly. Cun-es 1-5, 0, 1.0, 2.0. 3.0. and 4.0 /Â¿MCP1. respectively, ÃŸ.glutamate- (averages of two determinations each). dependent assembly. Cunes 1-5, 0, 1.5, 2.0, 3.0, and 4.0 /Â¿MCPI. respectively. GTP. All agents that bind in the Vinca domain also inhibit nucleotide exchange on tubulin, with VLB least active and maytansine. spongistatin I. DIO, and phomopsin A most active (2). When added cells. With CPI, we twice obtained an IC50 of 20 pM. The dolastatins prior to radiolabeled GTP or GDP, these agents inhibit nucleotide yielded values comparable to those obtained earlier, as did repeat binding to tubulin: but. as has been shown when tubulin with evaluation of spongistatin I, which had previously yielded IC,()s of [8-'4C]GDP bound in the exchangeable site is used, nucleotide bound 20-30 pM(17. 18).Thus, CPI is at least 15-fold more active than DIO in the exchangeable site is not actually displaced by any of the drugs and 150-fold more active than D15 against L1210 cells and compa (10,23). rable in its activity to the macrocyclic polyether. With CPI, the Initial studies showed that CPI also inhibited nucleotide exchange. mitotic index increased concordant with inhibition of cell growth, thus However, CPI appeared to be relatively inactive as an inhibitor of [8-'4C]GTP binding to tubulin (Fig. 4, â€¢¿),withan IC50 of about 50 confirming the report of Smith et al. (3) that CPI is an antimitotic depsipeptide of extraordinary potency. fjiM (versus about 11 Â¡JLMforDIO; Fig. 4, D). The activity of CPI

CPI Inhibits the Assembly of Purified Tubulin Table 1 Inhibition of lubnlin pofynuri&tion h\ CPI ami other VÃ¬ncatlomuin antimitotic dntgf Although the observation of potent antimitotic activity in a natural Inhibition of tubulin polymerization product strongly indicates an interaction with tubulin (2). other mech Drug IC50 (M.M)Â±SD anisms of action are feasible. Moreover, both inhibitory and stimula CPI 2.7 Â±0.2 tory effects on assembly reactions occur with agents that interact with DIO 1.8 Â±0.1 DIS 19 Â±0.3 tubulin. Furthermore, there is only a limited quantitative correlation VLB 3.9 Â±0.4 with different drug classes between relative antiproliferative activities Maylansine 4.5 Â±0.2 and effects on tubulin polymerization/1 One of the most striking "The concentration of drug required to inhibit glutamate-induced assembly of 1.0 nig/ml (10 /AM)purified tuhulin was determined, as described in the text. The extent of discrepancies is the low inhibitory activity of D15 on tubulin assem assembly after 20 min at 37Â°Cwas determined. A minimum of three independent bly, despite its potent inhibition of cell growth (15). experiments was performed with each drug. When we evaluated the effects of low concentrations (substoichio metric to the tubulin concentration) of CPI on tubulin assembly, we Table 2 Inhibition ofTHJVLS and Â¡3H]DIObinding tu tuhulin by CPI ami other Vinca domain drug*'1 observed progressive inhibition of assembly induced by both MAPs (Fig. 2A) and glutamate (Fig. 2B). In the MAP-dependent reaction, 'i inhibition about 50% inhibition of net assembly was observed with 2 Â¡J.MCPI, DrugCPIDIO(l9aÃ„)-Isodolaslalin binding333231343823DIO binding33292620001 using 15 /J.Mtubulin. The glutamate-dependent assembly reaction, using IO IJL.Mtubulin.was studied in greater detail to obtain quantita IOMaytansineHalichondrin tive comparisons of CPI with DIO and D15. as well as with VLB and maytansine (Table 1). Only DIO was more effective than CPI (IC50 of BRhizoxinD15VLBVLB 1.8 (LIMversus 2.7 JU.MforCPI). D15 was a relatively weak inhibitor of the assembly reaction (IC5(), 19 /J.M),whereas VLB and maytansine were moderately less effective than CPI. These results differ little "The 0.4-ml reaction mixtures contained 10 /IM (1.0 mg/ml) tubulin. O.I M MES (pH 6.9). 0.5 IHMMgCU \% DMSO. either [3H]VLB or [3H]D10 at 10 MM.and the indicated from those of Kerksiek et al. (6) and Smith and Zhang (7). inhibitor at 5.0 Â¿Â¿Mandwere incubated for 30 min at room temperature. Duplicate aliquots (0.19 ml) were placed on microcolumns of Sephadex G-50 (superfine) and were processed by centrifugal gel filtration at room temperature. Averages from two independent exper 6 However, often there is reasonably good correlation between antiproliferative activity iments are presented in the table. Stoichiometry of binding in the control reaction mixtures anil inhibitory effects on tubulin assembly within a specific class of drug. without inhibitor was 0.53 mol of VLB and 0.55 mol of DIO per mol of tubulin 44(10

spongistatin 1-9 (18), CPI inhibited turbidity development induced by DIO. Since CPI appears to bind at the same site on tubulin as DIO, we examined in greater detail the effects of CPI on DIO-induced aggregation. The time course of DIO-induced aggregation could be examined by turbidimetry, and Fig. 5 presents two studies with tubulin at 10 /LIM (1.0 mg/ml). The study shown in the main panel was performed at 0Â°C and that in the inset at 37Â°C.At 0Â°C,there was no turbidity change with tubulin alone (Fig. 5, curve I) or when the protein was mixed with 10-40 MMCPI (Fig. 5, cune 2). With DIO, the rate of aggre gation, but not its extent, depended on drug concentration. With 10 -4-2024 H.M DIO, there was no apparent reaction for 6 min, followed by sigmoidal kinetics, with turbidity approaching a plateau after 75 min MM VINBLASTINE (S) (Fig. 5, curve 3). This reaction was strongly inhibited by low concen trations of CPI (2 /UM,Fig. 5, curve 6; 5 /U.M,Fig. 5, curve 7). With 20 B 2.0 /MMDIO (Fig. 5, curve 4). following a I-min lag. a rapid increase in turbidity occurred. With 40 JU.MDU)(Fig. 5, curve 5), turbidity rose 1.5 rapidly as soon as the drug and tubulin were mixed. Raising the temperature of these reaction mixtures resulted in small decreases (10-20%) in the turbidity readings (data not shown), with no change 1.0 in turbidity in the sample containing 40 /Â¿MCP1. Because DIO-induced turbidity decreased at 37Â°Cand because we 0.5- were concerned that tubulin decay might have prevented a tempera-

MM DOLASTATIN 10 (S) 100- Fig. 3. Inhibition of ['H]VLB (Ai and ['H|D10 (B) binding to tubulin by CP1. Reaction mixtures (0.40 ml) contained 5.0 /AM(0.5 mg/ml) tubulin. 0.1 M MES (pH 6.9). 0.5 mM 80- MgCl,. 27r DMSO. the indicated concentrations of |'H|VLB or ['H]DIO, and the concentrations of CPI indicated below. Tubulin was the last component added to the reaction mixtures. After 30 min al room temperature. 0.2-ml duplicate aliquots of each c o 60- reaction mixture were processed by centrifugal gel filtration at room temperature. In A. the O following concentrations of CPI were used: â€¢¿.none;â€¢¿.2.0/AM:A. 3.0 /AM;and V. 5.0 /AM.Ordinate units for S/V: /AMVLB X /ig tubulin/pmol VLB bound. In B, the following concentrations of CPI were used: â€¢¿none:â€¢¿,2.0/AM:A. 4.0 fini: and T. 6.0 /AM.Ordinate 40- units for S/V: Â¡atDIO X /Â¿gtubulin/pmol DIO bound.

20- relative to that of DIO was dramatically enhanced by preincubating v. the drug with tubulin at 0Â°Cprior to the addition of [8-'4C]GTP to the reaction mixture (Fig. 4). When the drugs were added 60 min prior to the nucleotide, the IC50 for CPI was about 8 /MMand that for DIO was 10 20 30 40 50 60 about 5 /Â¿M(extrapolation from the data of Fig. 4). As with the other of Inhibitor Vinca domain drugs, CPI did not displace radiolabel from tubulin- [8-'4C]GDP and inhibited displacement of the bound [8-14C]GDP by Fig. 4. Effect of a 0Â°Cpreincubation of drug with tubulin on subsequent inhibition of the binding of [8-'4C]GTP to tubulin: comparison of CPI (â€¢.â€¢¿,A.and T) with DIO (D, nonradiolabeled GTP (data not presented). Smith and Zhang (7) found O, A, and V). Reaction mixtures (0.40 ml) contained IO /AM(l.O mg/ml) tubulin. 0.1 M that CPI inhibited tubulin-dependent GTP hydrolysis, but they were MES (pH 6.9). 0.5 mM MgCU. 1% DMSO. and the indicated drug concentration. Following preincubation on ice (â€¢and D. no preincubation: â€¢¿andO, 15 min preincu unable to demonstrate an effect on nucleotide exchange. However, bation: A and A, 30 min preincubation: V and V, 60 min preincubation). 50 /AM these workers did not evaluate CPI concentrations greater than 20 /Â¿M [8-l4C]GTP were added to each reaction mixture. After an additional 15 min on ice. or a drug-tubulin preincubation, which we found were necessary to 0.2-ml duplicate aliquots of each reaction mixture were processed by centrifugal gel filtration at 4Â°C. observe maximal effects on the nucleotide binding reaction. CLC. CPI did not significantly inhibit the binding of CLC to Table 3 Stabilisation of CLC binding activity of Inbitlin b\ CPI and other Vinca tubulin, as was also observed by others (6, 7). Under some reaction domain tÃ®ntes" conditions, most Vinca domain drugs enhance the binding of CLC to pmol [JH]CLC bound per pmol tubulin tubulin, and many of these drugs retard or arrest decay of the CLC binding activity of tubulin (2, 10, 24). CPI closely resembles DIO DrugNone preincubated0.31 both in preventing loss of CLC binding activity in a preincubation, compared with the partially protective VLB and the nonprotective CPI 0.53 0.510.52 DIO 0.530.46 maytansine, and in strongly enhancing CLC binding without a prein VLB 0.41 cubation, compared with the weaker effects of VLB and maytansine MaytansineNot 0.39Preincubated0.18 0.19 (Table 3). " Each 0.1-ml reaction mixture contained 0.4 mg/ml tubulin. 0.1 M MES (pH 6.4), O.I M EDTA. 1 mM GTP. 0.5 m.MMgCK. I mM 2-mercaptoethanol. 1 mM EGTA. 5% DMSO. 60 /AM[ HJCLC. and the indicated Vinca domain drug at 50 /ÃŒM.Ifindicated, the reaction Tubulin Aggregation Induced by CPI mixtures were preincubated for 3 h at 37Â°Cprior to the addition of the [~'H]CLC in lO/il. Following the addition of |'H|CLC, incubation was for 2 h at 37Â°C.The data presented In the above experiments, we never observed that reaction mixtures in the table represent average values obtained in two independent experiments, each of containing CPI became turbid. In fact, as in earlier studies with which contained triplicate samples. 4401

with 10 JUMDIO was completely inhibited by 2 /U.MCPI (Fig. 5. cun-e 4). It appeared that CPI was like maytansine. rhizoxin. and the macrocyclic polyethers and did not cause tuhulin aggregation and inhib ited the aggregation reaction induced by DK) (2). Even very low 1.0 concentrations of DK) relative to tubulin (as low as 1:25) caused formation of small oligomers detectable by si/e exclusion HPLC (13). With |'H]D10, we could not document binding of drug to single tubulin molecules. The smallest radiolabeled species comigrated with < a Mr 2(X),()00 standard and was distinct from the main protein peak. < We decided to determine whether CPI would inhibit formation of such small DIO-induced oligomers concordant with its inhibition of 0.5 the binding of ['H]D10 to tubulin. However. HPLC experiments with CPI demonstrated that such a study was not possible. CPI also caused formation of tubulin oligomers detectable by the HPLC technique. Quantitatively, the effects of D10 and CPI on oligomer formation were virtually identical, with stoichiometric amounts of either drug shifting the tubulin from the included volume to the void volume (Fig. 30 6). A comparable effect of stoichiometric CPI was observed by MINUTES Kerksiek el til. (6). Fig. 5. Comparison of the effects of ('PI and DK) in causing turbidity development in tuhulin solutions at ()Â°C(rnnin Â¡nini'l)and 37Â°C(/Vi.vt7).Eaeh 0.25-ml reaction mixture Morphological Studies contained 10 Â¿IM(1.0mg/ml) lunulin. 0.1 M MES (pH 6.9), 0.5 HIMMgCI,. 49f DMSO. and further components as indicated. Main panel: curve 1. no further addition; cun-e 2, 40 We turned to electron microscopy for information about aggregates MMCPI ( 10 and 20 Â¿IMCP1also had no effect): t iin-c.). 10 (Â¿MDIO;curve 4, 20 JIMD10: induced by CPI. Abundant ring-like aggregates were formed when cun-e 5. 40 JIM DIO; cun-e 6. K) JIM DK) and 2.0 JIM CPI; curve 7, 10 Â¡IMDK) and 5.0 fiM CPI. /Â».ve/;cun-e 1. no further addition; cun-e 2, 40 Â¿IMCP!( 10 and 20 ^M CPI also CPI was added to tubulin. and we observed little variation in their had no effect); cun-e 3. 10 (ÃŒMDIO:cun-e 4. IO P.M DK) and 2.0 JIM CPI. In the morphology under different conditions. No specific effects could be experiment shown in the inset, the temperature was jumped from 0Â°Clo 37Â°Cwithin 2 ascribed to reaction temperature, protein or drug concentration, GTP, min of adding DIO to the reaction mixtures. or MAPs. The morphology of the CPI-induced aggregate was distinct from that formed with either D10 or VLB. Because of the competitive ture-dependent aggregation reaction induced by CPI, we performed inhibition of DIO binding (versus noncompetitive inhibition of VLB the experiment shown in the Fig. 5 inset. The temperature of the binding), we performed comparative studies with CPI and DIO (a reaction mixtures was jumped to 37Â°Cwithin 2 min of adding DIO. comparative study between VLB and DK) was presented previously: Again, no change in turbidity occurred with tubulin alone (Fig. 5, see Ref. 13). cun'e 1) or when the protein was mixed with 10-40 JU.MCPI (Fig. 5, Fig. 7 compares high power views of tubulin aggregates in the cun'e 2). With 10 Â¿Â¿MDIO(Fig. 5, cun-e 3). following a lag of 2.5 absence of MAPs induced by CPI (Fig. 7A) with those induced by min, there was a sigmoidul rise in turbidity, with the plateau reached DK) (Fig. IB). With D10. the entire grid appeared to be covered by after 20 min. The turbidity reading was about one-half that observed rings, conjoined rings, and broken rings, as observed previously (13). at 0Â°C,but when the reaction temperature was returned to ()Â°C,there With CPI, there were clumps that had the appearance of multiple was only a small increase in turbidity (10-20%). The 37Â°Creaction clustered rings, together with scattered individual rings between

Fig. 6. Drug-induced aggregation of luhulin. Reaction mixtures contained 2.5 Â¿AM(0.25mg/ml) tuhulin, O.I M MES (pH 6.9). 0.5 nui MgCI,, 17t DMSO, and drug as indicated. Samples were Â¡ncuhated IO min at room tem perature prior to injection onto the HPI.C column. Injec tion volume was 0.5 ml. and flow rate was 1.0 ml/min. HPLC was performed at room temperature. A. no drug. B-F. CPI at 0.1, 0.25. 0.5. 1.0, and 5.0 Â¿Â¿M,respectively. C-K. DIO at O.I. 0.25. 0.5, 1.0. and 5.0 JXM,respectively. Arrows in Â£.F. ./. and K indicate the expected position of the A/, KX).(XK)tuhulin peak.

4402

Fig. 7. Comparison of the morphology of tubulin aggregates induced by CPI (A} and DIO (fl) in the absence of MAPs. X73.000. Reaction mixtures contained 10 (tM (1.0 mg/ml) tubulin, 0.1 M MES (pH 6.9). 0.5 mM MgCK, 4% DMSO. und the indicated drug at 20 /IM. Grids were prepared following a 40-min incubation at 37Â°C.

clumps. The CPI rings had a diameter about one-third to one-half that tially smaller on average than the clumps observed in the CPI sam of the DIO rings. VLB-induced aggregates had the appearance of ples. Invariably, at high power, the clumps from the tubulin-only ill-defined, short spirals (13). controls have not shown any discrete structural morphology (Fig. 9ÃŸ, Fig. 8 compares high power views of aggregates formed in the inset). presence of MAPs (panel A with CPI, panel B with DIO). As before (13), the MAPs seemed to impose a greater "order" in the appearance of the DIO-induced aggregate, with most of the material having the DISCUSSION appearance of conjoined rings. These structures probably represent The studies presented here, together with those of Kerksiek et al. tightly coiled filaments that sometimes take on the appearance of pinwheels,7 with the field shown in Fig. 8B displaying more coils and (6) and Smith and Zhang (7). establish that the depsipeptide CPI interacts strongly with tubulin, inhibiting microtubule assembly. This pinwheel formations than generally observed in a single field (the interaction probably accounts for the antimitotic activity of the drug view was chosen for the variety of structures shown in a small area). and for the disappearance of intracellular microtubules (3). In studies Previously (13), we had observed that MAPs also altered the mor of the effects of CPI on interactions of tubulin with other ligands, we phology of the VLB-induced aggregate, resulting in formation of the found a noncompetitive pattern of inhibition versus [3H]VLB, a looser spiral filaments described by many workers (reviewed in Ref. competitive pattern versus [3H]D10, inhibition of nucleotide exchange 26). CPI differed from DIO and VLB in that MAPs had little effect on that was enhanced by preincubating drug and tubulin, and strong the morphology of the aggregate formed in terms of its electron stabilization of the [3H]CLC binding activity of tubulin. Except for microscopic appearance. As in the absence of MAPs (Fig. 7/4), with enhancement of the inhibitory effect on nucleotide exchange by the MAPs (Fig. 8/4) the CPI aggregate largely consisted of conjoined drug-tubulin preincubation, these properties are identical to those of miniature rings. We have not visualized any well-defined coils or spirals in CPI-treated samples. the peptides DIO and phomopsin A. To account for these properties (especially the noncompetitive pattern of inhibition of Vinca alkaloid In view of the aggregated conjoined rings visualized by electron binding), as well as structure-activity observations with DIO ana microscopy, it seemed strange that CPI failed to cause a change in logues, we proposed that DIO and phomopsin A bind to a "peptide turbidity when added to tubulin with or without MAPs. We, therefore, site" on ÃŸ-tubulindistinct from the "Vinca site." This model proposed compared the overall low power appearance of tubulin + CPI to that inhibition of VLB binding and nucleotide exchange occurred tubulin alone (Fig. 9). With drug, there were many electron dense because occupancy of the peptide site sterically obstructed access to clumps of varying sizes. One of the largest seen is shown in Fig. 9/4. the Vinca and nucleotide sites (10). We suggested that this complex of Invariably, at high power, such clumps had the fine structure of ligand binding sites be called the "Vinca domain."8 Our ligand data clustered small rings (Fig. 9/4, inset) that results from the interaction support the idea that the depsipeptide CPI also binds in the peptide of CPI with tubulin. When scanning grids prepared from reaction site and not in the Vinca site itself. mixtures containing tubulin but no drug, we also observed many electron dense clumps (Fig. 90). Thus far, these have been substan * There is probably a third drug binding region in the Vinca domain that forms the primary binding site for the macrocyclic polyethers. because halichondrin B noncompeti- 7 Aggregate induced by the cyclic peptide phomopsin A is identical in appearance to tively inhibits the binding of ['HJVLB to tubulin ( 16) and spongistatin 1 noncompetitively that induced by DIO (37). inhibits the binding of both | *H]VLB and [-'HJDIO to tubulin (18). 4403

Fig. 8. Comparison of the morphology of tubulin aggregates induced by CPI (Ai and DIO (B) in the presence of MAPs. X73.000. Reaction mixtures contained components as described in the legend for Fig. 7 and heat-treated MAPs at 0.5 mg/ml and were incubated as described for Fig. 7. In B. the smaller arrow points to an aggregate rich in coiled structures, and the Itirxer urnw points to a pinwheel aggregate.

The major difference between CPI and DIO is in the structured the Vinca domain class of drugs to postulate a relationship between aggregates they induce. The drugs are quantitatively similar in the drug stabilization of CLC binding activity and drug induction of formation of small tubulin oligomers. These HPLC studies were structured aggregates. Thus far, every agent that causes aggregation performed on columns not equilibrated with free drug, indicating that (VLB, DIO, phomopsin A, and CPI) prevents the time-dependent CPI, like DIO (13), binds tightly to tubulin in a slowly reversible decay of the CLC binding activity of tubulin; whereas all agents that reaction. Yet DIO causes tubulin solutions to become highly turbid, do not induce aggregation (maytansine, rhizoxin, halichondrin B, whereas no change in turbidity occurs with CPI. Electron microscopy spongistatin 1, and D15) fail to stabilize the CLC binding activity of adds to the apparent paradox. Both drugs cause formation of substan tubulin. Only agents that constrain lubulin conformation within mul- tial amounts of structured aggregates with distinct morphology. The timeric aggregates appear to preserve the conformation of the CLC morphologies of these aggregates are different from each other, as site. well as from aggregates induced by VLB. The gross morphology of This generalization does not apply to the exchangeable nucleotide the CPI-induced aggregate is not greatly affected by MAPs, in con site. The preincubation effect resulting in enhancement of inhibition trast to substantial changes observed in the DIO- and VLB-induced of nucleotide exchange by CPI could result from slow aggregation, aggregates. These differences in drug effects probably result from but we have no method for evaluating the time course of CPI-induced subtle changes in tubulin-tubulin interactions as a function of the tubulin aggregation. More importantly, of the four most potent inhib bound ligand. itors of nucleotide exchange [maytansine (23), DIO (10), phomopsin What accounts for the limited ability of the CPI aggregate to scatter A (10), and spongistatin 1 (17)], only the peptides cause tubulin light? Two possibilities merit consideration. There appears to be some aggregation, and VLB is the weakest inhibitor of nucleotide exchange aggregate in purified tubulin preparations, as exemplified by the small among the Vinca domain drugs, despite its induction of an aggregation void volume peak observed in the HPLC studies without drug (Fig. reaction (16). Thus, we continue to favor steric inhibition, with drugs 6A; Ref. 6) and the unstructured, electron dense clumps observed by differentially affecting egress and entry of nucleotide at the exchange electron microscopy. Although such aggregates probably represent able site, as the most probable explanation for inhibition of nucleotide denatured tubulin, the large structured CPI-induced aggregates could exchange by this class of drugs. However, neither allosteric effects derive primarily from preexisting aggregates. The HPLC studies argue nor aggregation effects can yet be excluded as factors causing this against this explanation, unless soluble Mr 100.000 tubulin dimers phenomenon. Also, there has been excellent qualitative and quantita tive correlation between a drug's inhibition of nucleotide exchange only enter relatively small oligomers upon addition of CPI to the reaction mixture. Alternatively, the absence of turbidity could result and its ability to inhibit formation of a cross-link between cysteine-12 from the fine structure of the aggregate, which appears to consist of and cysteine-201 or -211 of ÃŸ-tubulinby /V,iV'-ethylenebis(iodoacet- relatively small interconnecting rings, possibly with an underlying amide) (12, 27, 28). CPI also inhibits formation of this intra-ÃŸ-tubulin tight spiral structure. cross-link (29). Enough data has now been accumulated with different members of Despite the quantitative similarity between DIO and CPI in most of 4404

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1996 American Association for Cancer Research. INTHRACTION OF CRYPTOPHYC1N 1 WITH TUBI LIN * â€¢¿*â€¢ B â€¢¿m *' -M â€¢¿&â€¢¿Â» * v , Â« Â« . ^ *

0- ^ -4^ Fig. 9. Comparison of the structured electron-dense aggregate observed in the presence of CP1 (Ai with the nonstructured aggregate observed in the absence of drug (ÃŸ).X4500 (main panels}; X224,000 (Â¡nwis). Reaction mixtures contained components as described in the legend for Fig. 7 and were incubated as described for Fig. 7. In each main panel, the arrow points to the region of the aggregate that is shown at high power.

their interactions with tubulin, the depsipeptide is much more potent a tubulin:drug ratio of 20:1.9 Our results thus far with [3H]D10 at the as an inhibitor of cell growth. With L1210 cells, CP1 was at least IC5(1concentration have been uninformative because of the relatively 15-fold more potent than DIO and had activity comparable to that of low specific activity of the peptide. Nonetheless, we tentatively con spongistatin 1. In our spongistatin 1 studies (17), we had compared clude that DIO does enter cells more readily than VLB, and we equivalent antiproliferative doses (IX and 10X the IC50 concentra postulate that CP1 will be even more readily absorbed. tions) of spongistatin 1, halichondrin B, DIO, CLC, and VLB on Assuming the low stoichiometry observed with VLB at the IC5(1 microtubules in PtKl cells. We observed similar effects with all concentration occurs universally with antimitotic agents that interact compounds, consistent with a similar mechanism of action for all with tubulin, what accounts for their potent action in cells? At least agents despite a 400-fold range in IC50s. Similarly, Smith et al. (3) two potential explanations merit consideration. The first is that sub- observed identical effects on microtubules of A-10 cells of CP1 and stoichiometric drug concentrations have significant kinetic effects on VLB at equivalent antiproliferative concentrations (100X the IC50 the functioning of cellular microtubules, probably by interfering with concentrations, which differed 500-fold). the dynamic instability properties of microtubules (31). The second is Little data is available on intracellular concentrations of antimicro- that y-tubulin is also a target of antimicrotubule agents, especially tubule drugs. Singer and Himes (30) measured VLB uptake in B16 those that inhibit assembly. Interference with nucleation at the micro- melanoma cells by an HPLC technique at 25 times and 250 times the tubule organizing center/centrosome via binding to y-tubulin (32-34) IC50 concentration and estimated by extrapolation that at that IC50 could readily account for the disruptive effects that even IC50 con concentration, the intracellular tubulin:drug ratio would be about centrations of antimitotic drugs cause on cellular microtubule organi 100:1. In attempting to gain insight into the greater antiproliferative zation (17). activity of DIO relative to VLB, we have compared intracellular Finally, although CP1 and not DIO is the central topic of this report, concentrations of radiolabeled drug in CA46 Burkitt lymphoma cells. our comparative studies with the two peptides have led to new At the same drug concentration, significantly more DIO than VLB enters cells. In studies at the IC50 concentration of VLB, we obtained ' A. Blokhin and E. Hamel, unpublished observations. 4405

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1996 American Association for Cancer Research. INTKRACTION OF CRYPTOPHYCIN I WITH TUBULIN observations on the tubulin aggregation reactions caused by the ma 13. Bai. R.. Taylor. G. F.. Schmidt. J. M.. Williams. M. D.. Kepler. J. A., Pettit. G. R., rine peptide. Most notable is the much greater turbidity development and Hamel. E. Interaction of dolastatin 10 with tubulin: induction of aggregation and that occurs at 0Â°Cas compared with 37Â°C.The temperature effect binding and dissociation reactions. Mol. Pharmacol.. 47: 965-976. 1995. 14. Pellit, G. R.. Srirangam. J. K.. Barkoczy. J.. Williams. M. D.. Durkin. K. P. M.. Boyd. does not appear to be reversible. After a prolonged incubation at 0Â°C, M. R.. Bai. R.. Hamel. E.. Schmidt. J. M.. and Chapuis, J-C. Synthesis of dolastatin the turbidity did not fall greatly when the reaction temperature was 10 structural modifications. Anti-cancer Drug Design. 10: 529-544. 1995. increased: and following a prolonged incubation at 37Â°C,the turbidity 15. Bai. R.. Friedman, S. J., Pettit. G. R.. and Hamel. E. Dolastatin 15. a potent antimilolic depsipeptide derived from Diilabella auricularia: interaction with tuhulin and effects did not rise greatly when the reaction temperature was decreased. on cellular microtubules. Biochem. Pharmacol.. 43: 2637-2645, 1992. Despite the different turbidity readings, electron microscopy did not 16. Bai. R.. Paull. K. D.. Herald. C. L., Malspeis, L.. Pettit, G. R.. and Hamel, E. Halichondrin B and homohalichondrin B, murine natural products binding in the reveal any obvious structural difference in aggregates formed at 0Â°C Vinca domain of tubulin: discovery of tubulin-bascd mechanism of action by analysis and 37Â°C. of differential cytotoxicity data. J. Biol. Chem.. 266: I5882-I5889. 1991 Previously, we have not observed electron micrographie differences 17. Bai. R.. Cichacz. Z. A., Herald. C. L., Pettit, G. R.. and Hamel. E. Spongistatin 1: a in tubulin aggregates induced by VLB at 0Â°Cand 37Â°C(13). In highly cytotoxic, sponge-derived, marine natural product that inhibits mitosis, microtubule assembly, and the binding of vinhlastine to tubulin. Mol. Pharmacol.. 44: contrast, different, and reversible, time-resolved X-ray solution scat 757-766, 1993. tering patterns were observed with VLB-induced tubulin aggregates at 18. Bai. R.. Taylor. G. F.. Cichacz. Z. A.. Herald. C. L.. Kepler, J. A.. Pcltit. G. R., and Hamel, E. The spongistatins. potently cytotoxic inhibitors of tubulin polymerization, 0Â°Cand 37Â°Cby Hodgkinson et al. (35) and Nogales et al. (36). The bind in a distinct region of the Vinca domain. Biochemistry, 34: 9714-9719, 1995. pattern differences were interpreted as indicating different pitches in 19. Pettit, G. R., Singh. S. B., Hogan. F., Lloyd-Williams. P., Herald. D. L., Burkctt. D. D.. and Clewlow. P. J. The absolute configuration and synthesis of natural the drug-induced spiral filaments. These reports indicate that it should (-)-dolastatin 10. J. Am. Chem. Soc.. ///: 5463-5465. 1989. be worthwhile studying effects of DIO and CP1 by this technique. 20. Pettit, G. R.. Herald, D. L., Singh, S. B.. Thornton, T. J.. and Mullaney. J. T. Antineoplastic agents. 220. Synthesis of natural (-)-dolastatin 15. J. Am. Chem. Soc. 113: 6692-6693. 1991. ACKNOWLEDGMENTS 21. Pettit, G. R.. Singh. S. B., Hogan, F., and Burkett. D. D. Chiral modifications of dolastatin 10: the potent cytostatic peptide ( 19aA)-isodolastatin 10. J. Med. Chem.. We thank J. F. Endlich for assistance with the electron microscopy. 33: 3132-3133. 1990. 22. Hamel, E.. and Lin, C. M. Separation of active tubulin and microtubule-associated proteins by ultracentrifugation and isolation of a component causing the formation of REFERENCES microtubule bundles. Biochemistry. 2.?: 4173-4184, 1984. 23. Huang, A. B., Lin, C. M.. and Hamel. E. Maytansine inhibits nucleotide binding at the 1. Weisenberg, R. C., Borisy. G. G., and Taylor. E. W. The colchicine-binding protein exchangeable site of tubulin. Biochem. Biophys. Res. Commun.. 128: 1239-1246. 1985. of mammalian brain and ils relation to microtubules. Biochemistry. 7: 4466-4477. 24. LudueÃ±a.R. F., Prasad. V.. Roach. M. C., and Lacey, E. The interaction of phomopsin 1968 A with bovine brain tubulin. Arch. Biochem. Biophys.. 272: 32-38. 1989. 2. Hamel. E. Antimitotic natural products and their interactions with tubulin. Med. Res. 25. Dixon, M., Webb, E. C., Thome. C. J. R.. and Tipion. K. F. Enzymes, pp. 332-381. Rev., 16: 207-231, 1996. New York: Academic Press. 1979. 3. Smith, C. D., Zhang, X.. Mooberry, S. L.. Patterson. G. M. L.. and Moore, R. E. 26. Hirnes. R. H. Interactions of the Calliaranthu.s (Vinca) alkaloids with tubulin and Cryptophycin: a new antimicrotuhule agent active against drug-resistant cells. Cancer microtubules. Pharmacol. Ther., 51: 257-267. 1991. Res., 54: 3779-3784. 1994. 27. LudueÃ±a.R. F.. and Roach. M. C. Tuhulin sulfhvdryl groups as probes and targets for 4. Barrow, R. A., Hemscheidl. T., Liang. J., Paik. S., Moore. R. E., and Tius, M. A. Total antimitotic and antimicrotubule agents. Pharmacol. Ther., 49: 133-152. 1991. synthesis of cryptophycins. Revision of the structures of cryptophycins A and C. J. 28. LudueÃ±a,R. F.. Roach, M. C., Prasad. V.. Pettit, G. R.. Cichacz. Z. A., and Herald. Am. Chem. Soc., 117: 2479-2490, 1995. C. L. Interaction of three sponge-derived macrocyclic lactone polyethers (spongistalin 5. Schwartz, R. E., Hirsch. C. F.. Sesin, D. F., Flor. J. E.. Chartrain. M., Fromtling, R. E., 3, halistatins 1 and 2) with tubulin. Drug Dev. Res., 35: 40-48, 1995. Harris. G. H.. Salvatore, M. J.. Liesch, J. M., and Yudin. K. Pharmaceuticals from cultured algae. J. Indus!. Microbiol., 5. 113-124. 1990. 29. Prasad. V.. Banerjee. M.. and LudueÃ±a.R. F. Interaction of cryptophycin with bovine brain tubulin. Proc. Am. Assoc. Cancer Res.. 37: 395. 1996. 6. Kerksiek. K.. Mejillano. M. R.. Schwartz. R. E.. Georg. G. I., and Hirnes. R. H. Interaction of cryptophycin 1 with tubulin and microtubules. FEBS Lett., 377: 59-61, 1995. 30. Singer, W. D.. and Himes. R. H. Cellular uptake and lubulin binding properties of four Vinca alkaloids. Biochem. Pharmacol., 43: 545-551, 1992. 7. Smith, C. D.. and Zhang. X. Mechanism of action of cryptophycin: interaction with the Vinca alkaloid domain of tubulin. J. Biol. Chem.. 271: 6192-6198, 1996. 31. Wilson. L.. and Jordan. M. A. Microtuhule dynamics: taking aim al a moving target. Chem. Biol., 2: 569-573, 1995. 8. Golakoti. T.. Ogino. J.. Heltzel. C. E., Husebo, T. L., Jensen, C. M., Larsen. L. K.. 32. Oakley, B. R., Oakley, C. E.. Yoon. Y.. and Jung. M. K. -y-Tubulin is a component Patterson, G. M. L., Moore, R. E., Mooberry, S. L.. CorbeÂ».T. H., and Valeriote. F. A. Structure determination, conformational analysis, chemical stability studies, and of the spindle pole body that is essential for microtuhule function in AspergiBus nidulans. Cell, 61: 1289-1301. 1990. antitumor evaluation of the cryptophycins. Isolation of 18 new analogs from Nostoc sp. strain CSV 224. J. Am. Chem. Soc., 117: 12030-12049, 1995. 33. Stearns. T.. Evans. L.. and Kirschner. M. y-Tubulin is a highly conserved component of the centrosome. Cell, 65: 825-836, 1991. 9. Bai. R.. Pettit, G. R.. and Hamel. E. Dolastatin IO. a powerful cytostatic peptide 34. Joshi, H. C.. Palacios, M. J.. McNamara. L.. and Cleveland. D. W. y-Tubulin is a derived from a marine animal: inhibition of tuhulin polymerization mediated through the Vinca alkaloid binding domain. Biochem. Pharmacol.. 39: 1941- centrosoma! protein required for cell cycle-dependent microtubule nucleation. Nature 1949. 1990. (Lond.). 356: 80-83. 1992. 10. Bai. R., Pettit. G. R., and Hamel. E. Binding of dolastatin 10 to tubulin at a distinct 35. Hodgkinson. J. L., Hutton. T.. Medrano. F. J.. and Bordas. J. X-ray solution scattering site for peptide antimitotic agents near the exchangeable nucleotide and VÃ¬nca studies of vinblastine-induced polymers of microtubule protein: structural characteri alkaloid sites. J. Biol. Chem., 265: 17141-17149. 1990. sation and effects of temperature. J. Struct. Biol.. 109: 28-38, 1992. 11. Bai. R.. Pettit. G. R., and Hamel. E. Structure-activity studies with chiral isomers and 36. Nogales. E., Medrano, F. J.. Diakun. G. P.. Mam, G. R.. Towns-Andrews, E., and with segments of the antimitotic marine peptide dolastatin 10. Biochem. Pharmacol.. Bordas. J. The effect of temperature on the structure of vinblastine-induced polymers 40: 1859-1864, 1990. of purified tuhulin: detection of a reversible conformational change. J. Mol. Biol., 12. Bai. R., Roach, M. C.. Jayaram, S. K.. Barkoczy, J., Pettit, G. R., Luduena, R. F., and 245: 416-430. 1995. Harnel. E. Differential effects of active isomers, segments, and analogs of dolastatin 37. Hamel. E. Natural products which interact with tubulin in the Vincti domain: rnay- 10 on ligand interactions with tubulin: correlation with cytotoxicity. Biochem. Phar tansine, rhizoxin. phomopsin A, dolastatins 10 and 15 and halichondrin B. Pharmacol. macol.. 45: 1503-1515. 1993. Ther., 55: 31-51, 1992.

4406

Downloaded from cancerres.aacrjournals.org on October 2, 2021. © 1996 American Association for Cancer Research. Characterization of the Interaction of Cryptophycin 1 with Tubulin: Binding in the Vinca Domain, Competitive Inhibition of Dolastatin 10 Binding, and an Unusual Aggregation Reaction

Ruoli Bai, Robert E. Schwartz, John A. Kepler, et al.

Cancer Res 1996;56:4398-4406.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/56/19/4398

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/56/19/4398. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.