Pharmac. Ther. Vol. 51, pp. 257-267, 1991 Printed in Great Britain. All rights reserved

0163-7258/91 $0.00 + 0.50 © 1991 Pergamon Press plc

Specialist Subject Editor: E. HAMEL

INTERACTIONS OF THE C A T H A R A N T H U S (VINCA) ALKALOIDS WITH TUBULIN A N D MICROTUBULES RICHARD H . HIMES Department o f Biochemistry, University o f Kansas, Lawrence, K S 66045-2106, U.S.A. AImtraet--The dimeric Vinca alkaloids represent a group of important anti-tumor compounds whose intracellular target is tubulin, the protein monomer of microtubules. In this review data on the binding of these drugs to tubulin and microtubules in vitro are examined. The binding to tubulin is linked to a protein self-association reaction described by Na and Timasheff (1986a) as a ligand-induced plus ligand-mediated isodesmic self-association reaction. The simplest model which fits the binding data is one in which there is one intrinsic site which is linked to the self-association process. Effects of solution variables on the binding and self-association explain the wide variation of reported apparent binding constants for Vinca alkaloids to tubulin. The Vinca drugs also bind to microtubules via a low number of sites at the ends of microtubules with apparent high affinity and which are involved in the inhibition of tubulin dimer addition to the microtubule ends, and to sites along the microtubule wall with apparent low affinity which are involved in the disruption of the microtubules into spiraled protofilaments. This review also compares available binding data for different natural and semi-synthetic Vinca alkaloids.

CONTENTS 1. Introduction 2. Tubulin Binding to Vinblastine 3. Relationship of Binding to Vinca Alkaloid-Induced Tubulin Aggregation 4. Binding of Different Vinca Alkaloids to Tubulin 5. Binding to Microtubules 6. Vinca Alkaloid-Induced Paracrystalline Structures 7. Conclusions Acknowledgements References

1. I N T R O D U C T I O N

257 257 259 260 262 265 266 266 266

(Wilson and Friedkin, 1967; Wilson, 1970), by the ability of these compounds to precipitate tubulin from cell extracts and from purified tubulin preparations (Marantz et al., 1969; Olmsted et al., 1970; Wilson et al., 1970; Weisenberg and Timasheff, 1970), and by the ability to induce the formation of paracrystaUine tubulin-alkaloid structures in vivo and in vitro (Schochet et al., 1968; Bensch et al., 1969; Bensch and Malawista, 1969; Stebbings, 1971; Bryan, 1971, 1972a). In this article I will review studies of the interactions o f the alkaloids with tubulin and microtubules in vitro. It would seem that the binding of an alkaloid to a protein should be a straightforward problem but the tubulin-Vinca alkaloid interaction has turned out to be quite complex.

Vinblastine (VLB) and vincristine (VCR) were first isolated about thirty years ago from the plant C a t h a ranthus roseus (L.) G. Don (formerly Vinca rosea), commonly known as Madagascar periwinkle (Noble et al., 1958a,b; Johnson et al., 1960, 1963; Cutts et al., 1960; Svoboda, 1961). Since that time innumerable articles describing the interactions of these and other naturally occurring and semi-synthetic Vinca alkaloids with tubulin and microtubules have been published. That the Vinca alkaloids interact with microtubules and tubulin was first indicated by the effects on the mitotic spindle in cultured cells (George et aL, 1965; Krishan, 1968), from the fact that they stabilize the colchicine binding property of tubulin without affecting the tubulin-colchicine reaction itself

2. T U B U L I N B I N D I N G T O V I N B L A S T I N E Abbreviations: VLB = vinblastine; VCR = vincristine; VRD = vinrosidine; VPD = vinepidine; VDS = vindesine; Pipes = piperazine - N,N" - bis(fl - aminoethylether)N,N'tetraacetic acid; MAPs = microtubule-associated proteins; MTP = microtubule protein.

In vitro binding studies have been performed using tubulin primarily from mammalian brain tissue but purified in different ways; by anion exchange chromatography and precipitation with M g 2+, by cycles of

257

R. H. HIMES

258

TABLE 1. Binding o f V L B to Tubulin Tubulin source

Number of sites

Solution conditions

Porcine brain Calf brain Chick brain Sea urchin sperm tail outer doublet Rat brain Calf brain Calf brain Bovine brain Bovine brain

10mM phosphate-100 mM glutamate,').l mM GTP, pH 6.5 l0 mM phosphateq). 1 mM GTP, pH 7.0 20 mM pyrophosphate-150 mM NaC1, pH 7.0 20 mM pyrophosphate-150 mM NaC1, pH 7.0 10 mM phosphate-10 mM MgCI2-0.1 mM GTP, pH 6.8 10 mu phosphateq). 1 mM GTP, pH 7.0 100 mM Mes-0.5 mM MgCI 2, pH 6.4 100 mM Pipes-I mM MgSO 41 mM EGTA, pH 6.9 10 mM phosphate, pH 7.0

~0.4

K, M ' × 10- 6 6

Reference Owellen et al., 1972

2.0

0.023

2.0

0.3qL5

Wilson et al., 1975

2.0

0.1~).3

Wilson et al., 1975

2.0

6.2 a 0.08 0.04

Bhattacharyya and Wolff, 1976

Safa et al., 1987

1.0

1.8a 0.071 2.5 5

Singer et al., 1988

1.0

0.02

Singer et al., 1988

1.0 2.0

Lee et al., 1975

Na and Timasheff, 1986a

aThe two values are for one high and one low affinity site. assembly a n d disassembly a n d by cycles of assembly-disassembly followed by cation exchange chrom a t o g r a p h y . In some cases the p r e p a r a t i o n s c o n t a i n e d significant a m o u n t s o f microtubule-associated proteins (MAPs). M e t h o d s used for studying binding have included filtration on D E A E filter p a p e r (Owellen et al., 1972; Wilson et al., 1975; Bhatt a c h a r y y a a n d Wolff, 1976), spectrofluorometry (Lee et al., 1975), b a t c h gel elution (Na a n d Timasheff, 1986a), a n d gel filtration (Wilson et al., 1975; Lee et al., 1975; Safa et al., 1987; Singer et al., 1988). In spite of these differences, the reported binding constants are quite similar when c o m p a r e d u n d e r nearly identical solution conditions. However, in examining the reported values it is readily evident that, when the solution conditions do differ the K~ values vary considerably. Table 1, which summarizes some data from the literature on VLB binding, illustrates this point. E x a m i n a t i o n of Table 1 shows t h a t reported K a values vary over a 310-fold range. In one study

bovine b r a i n tubulin p r e p a r e d by two methods, the assembly-disassembly m e t h o d a n d the m e t h o d using a n i o n exchange c h r o m a t o g r a p h y a n d M g 2+ precipitation, were c o m p a r e d (Singer et al., 1988). N o difference in binding affinity was seen when binding was studied in identical solutions but the affinity o f b o t h tubulin p r e p a r a t i o n s for VLB was greatly affected by the solution c o m p o n e n t s . It is clear from inspection o f Table 1 that ionic strength a n d M g 2+ have a large effect on the value of Ka, with higher values being observed in solutions of high ionic strength or high M g 2+ concentration. M o s t Scatchard plots for Vinca alkaloid binding are curvilinear a n d a p p e a r to fit a binding isotherm for one high affinity site and one or more lower affinity sites. The effect of solution variables on the Ka value of the a p p a r e n t high affinity site was examined in a systematic way a n d the results are presented in Table 2 (Singer et al., 1988). Increasing the Pipes c o n c e n t r a t i o n from 10 mM to 100 mM in P E M buffer caused a 17- to 25-fold increase in the K,. Inclusion

TABLE2. Effect o f Solution Variables on the Apparent High/{fit'nity Association Constant Jbr V L B

Buffera PEM, 0.01 M PEM, 0.1 M PEM, 0.1 M Pipes, 0.1 M Phosphate, 0.01 M Phosphate, 0.1 M Phosphate, 0.01 M MgCI2, 0.5 mM Phosphate, 0.01 M MgCI2, 10 mM

Ionic strength mM

[Tubulin], /iM

Metho&

28 219 219 212 18 180 20

10 2 0.5 2 10 10 10

1 2 3 I 1 1 2

0.2 3.4 5.0 0.7 0.02 1.0 0.32

48

2

1

1.4

K~, M-i

x 10

~

~Buffers: 0.1M PEMq).I M Pipes, pH 6.9, 1 mM EGTA, l mm MgSO4; 0.01M PEM-PEM with 0.01 u Pipes. The buffers containing Pipes were pH 6.9. The phosphate buffers were pH 7.0 and contained sodium. bBinding Assay Methods: 1. column centrifugation; 2. batch gel elution; 3. micropartition. All studies were done at 22°C.

Interactions of the catharanthus alkaloids of 1 m M Mg 2+ with 100 mM Pipes increased the value by 5- to 7-fold. Increasing the phosphate concentration from 10mM to 100mM increased Ka by 50fold. Finally, the addition of 0.5 mM Mg 2+ to 10 mM phosphate resulted in a 16-fold increase, while the addition of 10 mM Mg 2÷ raised the value by 70-fold. There was a 250-fold difference between the lowest and highest values observed in this study, similar to the differences found in the literature. Some of the agents that increase the apparent Ks value are also known for their ability to promote tubulin aggregation and polymerization, i.e. Mg 2÷, organosulfonates, the proper ionic strength.

3. RELATIONSHIP OF BINDING TO VINCA ALKALOID-INDUCED TUBULIN AGGREGATION The fact that VLB precipitates tubulin indicates that the alkaloid promotes aggregation of the protein. Early biochemical studies of tubulin confirmed this. Using analytical ultracentrifugation Weisenberg and Timasheff (1970) demonstrated not only that VLB caused the 6 S tubulin dimer to aggregate into species with sedimentation constants of up to 18 S, but that Mg 2÷ potentiated the aggregation process, causing the formation of still higher molecular weight species. Tubulin aggregation induced by VLB and VCR has been studied thoroughly by Timasheff and colleagues (Lee et al., 1975; Na and Timasheff, 1980a,b, 1986a,b; Prakash and Timasheff, 1985). In these studies sedimentation velocity determinations were performed as a function of protein, Vinca alkaloid and Mg 2÷ concentrations and were correlated with binding experiments. The results demonstrated that the curvilinear binding isotherms could be best fit by a system described as a one-ligand-induced isodesmic indefinite self-association. The best fit was a mechanism of association which involves both a ligand-mediated and ligand-facilitated pathway. A simplified version of this concept is given in Fig. 1. Several important points came from the studies in TimashelTs laboratory. One concerns the meaning of the experimentally determined association constant

259

for Vinca drug binding. It is an apparent constant, and as is evident from Fig. 1, there are several equilibrium constants involved in the production of tubulin-alkaloid aggregates. In 10ram phosphate buffer, a value of 4 x 104M -1 for Kl, the intrinsic binding constant to the self-association linked site, gave the best fit to the data (Na and Timasheff, 1986a). The self-association constant for the tubulin-Vinca alkaloid complex, K2, which fit the data, was 1.8 x 105 M-l, and the binding constant of VLB to polymerized tubulin, K3, was >_4 x 106 i -~. K4, the constant for tubulin self-association in the absence of bound alkaloid was small, < 2 x 103 i -1. In addition, weak binding sites, not linked to the self-association process, with a binding constant of _

6 4

~'**b

o o TU2 .

~

Tu2V 2

FIG. 1. Vinca drug ligand-induced and ligand-mediated association of tubulin. This is a simplified version of the model described by Na and Timasheff (1986a). It represents only the formation of a dimer of the tubulin-Vinca complex but indefinite aggregation takes place.

* *[°l°~ ~A,~&AA , 1 .o 2.0 VLBb/Tubulin

FIG. 2. Scatchard plot of VLB binding to tubulin. The protein concentration was 16 # Mand the buffers were I0 mM phosphate-0.1 mM GTP, pH 7, with (O) and without (A) 1 mM MgCI2. The data points were taken from Fig. 1 of Na and Timasheff (1986b).

260

R.H. Hit,ms

the discrepancies in reported binding constants found in the literature. The above discussion points to the fact that the binding of Vinca alkaloids to the tubulin molecule is linked to a self-association reaction and that M f ÷ affects the linkage. Mg 2+ itself, at high enough concentrations, promotes tubulin aggregation (Frigon and Timasheff, 1975a,b). Other tubulin assembly promoting agents, such as the organosulfonate buffers also apparently enhance the binding and self-association constants, and the linkage between binding and self-association. It has not been possible to separate binding from self-association; whenever Vinca alkaloid binding is detected, aggregation is also observed. In Figs 3 and 4 the degree of aggregation, detected by analytical ultracentrifugation and size exclusion-HPLC, is observed as a function of VLB concentration. The degree of aggregation in 10mM phosphate buffer, conditions where the observed binding constant is lowest, is much less than in Pipes buffer containing Mg 2+.

a

Vka.0

5-5.7

~1

b

VLB T~-=0".~

5-5.7

~-

C

VLB 1 0

5-13.4

[

E

e

Tu

~-z.5

J

S=7.3 J

Distance FIG. 3. VLB-induced aggregation of tubulin as detected by analytical ultracentrifugation using UV-optics. The initial VLB/tubulin molar ratios are given with the tubulin concentration constant at 5 tiM. The buffer was PEM in panels a~l, and 10 mr~ phosphate in panel e. The sedimentation coefficients given in the panels were calculated from the midpoint of the curves. Reprinted from Singer et al. (1988), with permission of the copyright holder, Pergamon Press plc, Oxford.

In summary, Vinca alkaloid binding to tubulin is a complex process of binding and self-association which produces large tubulin~lrug aggregates. The data fit a model in which there is one primary site which is linked to the aggregation process. Other, much weaker binding sites also exist. Both the intrinsic binding constant and the self-association constant are dependent on solution conditions and are increased by agents which promote tubulin aggregation. This makes it crucial that solution conditions, including protein concentration, be identical when comparisons are being made between laboratories or when different alkaloid drugs are being compared. It also makes it difficult to evaluate binding properties under in vivo conditions.

4. BINDING OF D I F F E R E N T VINCA ALKALOIDS TO TUBULIN Several different natural dimeric Vinca alkaloids have been isolated and numerous semisynthetic derivatives of them have been made, primarily by Eli Lilly & Co., Indianapolis, IN. The structures of five of these compounds are shown in Fig. 5. Chemical modifications have been made in the velbanamine portion, the upper portion in Fig. 5, and in the vindoline portion. Very little tubulin binding data have been reported on the many derivatives. In those cases where data are available and where comparisons have been made using the same solution conditions, it has been found that most changes in the molecule have only small effects on the binding. For example, in a recent study (W. D. Singer and R. H. Himes, unpublished results) it was found that the order of K, values was vinepidine (VPD) > VCR > vindesine(VDS) > VLB, but the difference in K, values for VPD and VLB was only four-fold. In other studies, in which different solution conditions were used, it was also found that VCR had a binding constant which was about 50% higher than for VLB (Owellen et al., 1972, 1976, 1977; Prakash and Timasheff, 1983). In another case, it was found that VCR had an affinity constant about four times that of VLB (Conrad et al., 1979). In several instances VDS showed a binding affinity somewhat stronger than that of VLB and slightly weaker than that of VCR (Owellen et al., 1976; Conrad et al., 1979). Thus, interchanging a formyl and methyl group at the N-I position, conversion of the methyl ester at C-3 to an amide, and deacetylation at C-4, have small effects on the binding affinity to tubulin. In fact, the addition of a rather bulky group, 4-azido-2-nitrophenyl-2aminoethyl, to the amide at C-3 of VDS has little effect on the binding properties of the alkaloid (Nasioulas et al., 1990). A related derivative with a 4-azido-benzoyl-2-aminoethyl group at the amide of VDS is reported to have an eight-fold lower binding affinity than VLB, even though its inhibiting effect on tubulin self-assembly is essentially identical to that of

Interactions of the eatharanthus alkaloids

b

8

d

C

VLB

VLB . "~--0.>

.

T~"0.1

VT'~"1.0

261

e

f

VT--2.

Vr -2.5

c

St

I

4

8

12

8 12 Elutiontime minutes

12 4

8

8

12

8

12 4

I

I

8

12

FIG. 4. VLB-induced aggregation of tubulin as detected by size exclusion HPLC. In panels a-e the tubulin concentration was 2 #M and 0.5/~M in panel f. The initial VLB/tubulin molar ratios are as shown. PEM buffer was used and the column eluting buffer contained the same concentration of VLB as the original solution to maintain equilibrium conditions. Reprinted from Singer et al. (1988), with permission of the copyright holder, Pergamon Press plc, Oxford.

VLB (Safa et al., 1987). The binding affinity of this latter derivative to tubulin in calf brain homogenates also appeared identical to that of pure tubulin (Safa and Felsted, 1987). Other derivatives which have been tested for their tubulin binding properties include VPD, a deoxy epimer at C-4' of VCR, and vinrosidine, (VRD) a C-4' epimer of VLB. VPD has an affinity about twice that for VCR (W. D. Singer and R. H. Himes, unpublished results) and VRD binds

I:15 ~

~

"

~

R4

i

A

R1

n2 R4

R5

Vincristine CHO COOCH30COCH 3

CH2CH3

OH

Vinblastine CH3

COOCH30COCH 3

CH2CH3

OH

Vindesine

CONH2

CH2CH3

OH

Vinepidine CHO COOCH30COCH 3

H

CH2CH ~

Vinrosidine CH3

OH

CH2CH..

R1

CH3

R2

P'3

OH

COOCH30COCH 3

F1G. 5. Structure of five Vinca alkaloids. Reprinted from Jordan et al. (1991), with permission of the authors and the copyright holder, American Association for Cancer Research, Philadelphia.

with about 20 to 50% the affinity of that of VLB (Owellen et al., 1977; Gerzon, 1980). In some cases Vinca derivatives have been compared in their effects on the rates of tubulin assembly or steady state dimer addition to the ends of microtubules. These assays reflect tubulin binding properties and the results are consistent with direct binding data. For example, in an assay which measures the rate of addition of tubulin dimers to microtubule ends VPD was the most effective inhibitor, VLB the least, with VDS and VCR intermediate between the two (Jordan et al., 1985). In the in vitro assembly assay small differences between the inhibitory activity of VLB, VCR, VRD, the 4'-deoxy derivatives of these compounds and anhydro VLB (VLB with a double bond between 3' and 4' and no hydroxyl at 4') were found (Zavala et al., 1978; Potier et al., 1979). However, the latter investigators found that removal of the methoxy carbonyl group from the C-18' of V L B or anhydro VLB led to a complete loss in inhibitory activity. Further, they showed that inversion of the configuration at C-18' or C-2' caused a large decrease in the activity of the alkaloids. In the few cases where it has been examined, the relative strength of binding has been found to be related to the aggregation-inducing ability of the alkaloids. Thus, Prakash and Timasheff (1985) reported that binding of VCR and the self-association reaction of the tubulin-VCR complex were stronger than the corresponding reactions for VLB. In another study, the degree of aggregation induced by four Vinca alkaloids and measured by size exclusionHPLC showed the same relative order as the K a values, i.e. V P D > V C R > V D S > V L B (W. D. Singer and R. H. Himes, unpublished results).

262

R.H. HIMES

The individual monomeric units of the dimeric Vinca alkaloid interact with tubulin much more weakly than the intact molecule. For example, vindoline and catharanthine (the latter is related to the velbanamine portion) show little or weak binding to tubulin (Owellen et al., 1977; Gerzon, 1980). These two compounds, as well as velbanamine, also show much decreased activity in inhibiting microtubule assembly (Prakash and Timasheff, 1991; M. A. Jordan and L. Wilson, personal communication). Binding and the induction of tubulin aggregation by catharanthine could be detected but the interaction with tubulin was about 10-fold weaker than that of VLB or VCR and the interaction with vindoline could barely be detected (Prakash and Timasheff, 1991). The calculated association constant for the addition of a tubulin-catharanthine complex to a growing polymer, however, was not much different from that for the tubulin-VLB or tubulin-VCR complexes. It appears, therefore, that once the compound is bound to tubulin, the thermodynamics of the association process is similar to that for the dimeric alkaloids.

5. B I N D I N G TO M I C R O T U B U L E S Are the high and low affinity Vinca binding sites in tubulin available to interact with the drug when the protein is incorporated i n t o the wall of a microtubule? This question has been addressed in several ways. Morphological evidence gives the impression that the alkaloids can bind to the walls of microtubules in vitro. Warfield and Bouck (1974) demonstrated that the addition of VLB to microtubules prepared from microtubule protein (tubulin plus associated proteins) resulted in the formation of helices composed of two protofilaments. These formed as a consequence of splaying at both ends of the microtubule (see Fig. 6 as an example). The authors felt that these spiral-like structures compacted and formed macrotubules, a commonly seen structure in cytosols after treatment of cells with a Vinca alkaloid. It has been proposed that such spiral formation of protofilaments constitutes the basic structure in the VLB-induced paracrystals (Fujiwara and Tilney, 1975). Vinca-induced protofilament spiral formation depends on the presence of MAPs bound to the microtubules (Donoso et al., 1979) or the presence of high Mg 2+ concentrations (Haskins et al., 1981). It appears that the two major brain MAPs, tau and MAP2, interact with tubulin somewhat differently to produce morphologically distinct spirals (Luduefia et al., 1981, 1984). It has been proposed that the MAPs stabilize the longitudinal interactions between dimers in the protofilaments preventing their complete disassembly as they splay apart after binding the Vinca alkaloid (Donoso et al., 1979). An illustration of this model is presented in Fig. 7.

ii! y!

¸

:Li

FIG. 6. Production of Vinca-induced spirals from microtubules. Microtubules formed in vitro from bovine brain MTP were treated with 200/~M VLB for l min. Magnification 130,000 ×. The micrograph was kindly donated by Dr. M. A. Jordan. The implication from the fact that VLB causes the conversion of microtubules to spiraled-protofilaments is that the drug binds along the wall of the microtubule, causing protofilaments to separate and subsequently coil. However, there are alternative explanations. Binding sites may only be available at the ends of the microtubules and as the ends splay apart previously buried sites become available, thus giving rise to a zipper effect. Another possibility is that spirals form from depolymerized protein since this is also known to occur (Warfield and Bouck, 1974; Erickson, 1975). A class of VLB binding sites with apparent low affinity but high capacity was identified on intact microtubules in vitro in the process of depolymerization (Jordan et al., 1986). An attempt was made to determine the binding affinity of these sites by stabilizing the microtubules against splaying with taxol and DMSO (Singer et al., 1989). The results indicated the presence of 1.4-1.7 weak binding sites per dimer

Interactions of the catharanthus alkaloids

263

FIG. 7. Model of the Vinca alkaloid-induced disassembly of MAP-containing microtubules into spiraled protofilaments containing one or two spirals. Reprinted from Donoso et al. (1979), with permission of the copyright holder, American Association for Cancer Research, Philadelphia.

with a Ka of 3-4 × 103M-1. Such weak sites had been hypothesized to be responsible for the splaying of protofilaments at relatively high Vinca drug concentrations (Jordan et al., 1986). The disintegration of the microtubules into protofilaments has been proposed to occur by a propagated mechanism, initially involving binding of the drug to a limited number of sites, causing weakening lateral interactions between protofilaments and resulting in the exposure of new sites (Jordan et al., 1986). In addition to the weak binding sites described above, VLB binds to a very small number of sites per microtubule (16.8 + 4.3 per microtubule out of a potential number of 17,000 tubulin dimers per average 10 # m microtubule) with a much higher apparent affinity (Ka = 5.3 x 105 M-~) (Wilson et al., 1982). The most likely location of these sites is at the ends of the microtubules. Inhibition of the assembly of microtubules shows substoichiometric poisoning. Using the concentration that produces 5 0 o inhibition together with the association constant of 2.3 × 105 M- 1one can calculate that 50% inhibition of the rate of assembly occurs when about 1 molecule of VLB is bound per microtubule (Wilson et aL, 1982). Kinetic studies of this inhibition demonstrated that the major effect of low concentrations of VLB on microtubules at steady state is to decrease the rate constants for dissociation and association of a tubulin dimer at the net assembly (A) end of the molecule, producing a kinetic cap (Jordan and Wilson, 1990). VLB also appears to decrease the association rate constant at the disassembly (D) end without affecting the D-end dissociation rate constant. The binding site with a Ka value of 5.3 × 105M-~ for binding of VLB to the end of a microtubule may be related to the isodesmic indefinite self-association model described by Timasheff's laboratory and discussed above. The value is 10-fold higher than the intrinsic binding constant to tubulin reported by Na

and Timasheff (1986a) but this constant could be affected by solution variables (see discussion above). In fact, in the presence of 1 mM Mg 2÷, which was used in the studies in Wilson's laboratory, Na and Timasheff estimated that the intrinsic binding constant could be 3.8 x 105M-I (Na and Timasheff, 1986b). In summary, intact microtubules appear to have at least two classes of Vinca alkaloid binding sites, one with apparent high affinity at the ends of microtubules, which could represent the intrinsic binding site on tubulin and is the site responsible for substoichiometric poisoning of tubulin assembly. The other class has a lower apparent binding constant and appears to be along the wall of the microtubule. The latter class appears to be responsible for the separation of protofilaments when microtubules are treated with high concentrations of VLB. The difference in apparent affinities for these two sites may be due to partial concealment of the intrinsic site in the microtubule wall or to a conformationai difference in the tubulin dimer in the core of the microtubule compared to the end of the microtubule. Borman and Kuehne (1989) examined three different activities of eight C-4' congeners of VLB: to inhibit microtubule assembly, disassemble preformed microtubules, and induce protofilament spiral formation. They found that the three activities were sensitive to changes at the C-4' site and that they were affected differently. 4'-Deoxy VLB and 4'-epideoxy VLB had activities similar to those of VLB. When the alkyl function at C-4' was removed, potency as an inhibitor of assembly was reduced 10-fold, single spirals as opposed to spiral aggregates were produced from soluble MTP, and normal spiral aggregates were produced from preformed microtubules. Replacement of the ethyl of the deoxy- and epideoxy VLB molecules with a methyl group gave rise to drugs which produced spiral structures during the

FIG. 8. Electronmicrographs of B16 m e l a n o m a t u m o r tissue incubated with V C R or VDS. T u m o r s were grown s.c. in mice, removed and treated with 50/~M o f the Vinca drugs for 2 hr. (A) Paracrystal structures (C) in different locations of the cytoplasm were identified in the tissue incubated with VCR. Some collagen (Co) can be seen between the cells. Magnification 9,000 x . (B) VDS-induced paracrystals (C). The presence of microfilaments (MF) were occasionally observed. Magnification 11,500 x . (C) Longitudinal section o f a paracrystal (C) induced by V C R adjacent to a large mitochondrion (M). Magnification 34,000 × . Unpublished results of J. A. D o n o s o and R. H. Himes.

264

Interactions of the catharanthus alkaloids assembly reaction at a concentration of 2 x 10-7 ~. Previously, spiraled protofilaments had not been observed at such low concentrations of a Vinca alkaloid. Three other 4'-congeners, desethyl VLB, dimethyl deoxy-desethyi VLB and e~pi-propyl deoxy-desethyl VLB, had ICso values for inhibition of microtubule assembly similar to that for deoxy-desethyl VLB but did not disassemble microtubules into spiraled protofilaments at 10-4M as did other Vinca congeners. From these studies it appears that the binding affinities of the various congeners to tubulin, MTP and microtubules may have been altered differently. Previously Zavala et al. (1978) and Potier (1979) had also demonstrated that a variety of derivatives of the Vinca alkaloids had different effects on the assembly reaction and the formation of spirals from microtubules. It would be informative to know the apparent binding constants of these various compounds for tubulin and the effects on the Vinca-induced tubulin self-association reaction.

6. VINCA ALKALOID-INDUCED PARACRYSTALLINE STRUCTURES Solution conditions not only influence the apparent binding constant for Vinca alkaloids and the degree of protein association, but also the structure of the

ee=

I

• ,,

265

polymerized product. As pointed out in the Introduction, early studies with cells in culture demonstrated that the Vinca drugs, at high concentrations, produce crystalline-like structures in cells. Examples of crystals induced by VCR and VDS in B16 melanoma tumor tissue are presented in Fig. 8. VLB crystals induced in sea urchin eggs contain one bound VLB per tubulin dimer (Bryan, 1972b; Wilson et al., 1978). Bryan (1972b) reported that such crystals could not bind additional VLB but a later report presented evidence for an additional VLB site (Wilson et aL, 1978). Both sites had an apparent Ka of 2.4 x l05 M-~. VLB-induced paracrystals have also been formed in vitro from calf brain tubulin (Na and Timasheff, 1982). These paracrystals contained about 1 VLB/dimer when 35 #M tubulin and initial concentrations of VLB from 40 to 300 #M were used. At concentrations higher than 300 #M the drug content of the paracrystals increased, eventually to 1.8 VLB/dimer when 1 mM VLB was used. The second site appears to be of low affinity and not associated with the crystallization process (Na and Timasheff, 1982). From both the in vivo and in vitro studies it is apparent that the binding of only one Vinca alkaloid per dimer is required to produce the paracrystalline structure, most likely at the same site which is linked to the protein self-association process. X-ray diffrac-

p=C

e° I

i

~

,

p=2c

i ;

eoe e I - - "

m~

""-i

•."7 J • °e

,~

--"

J

..'"-i',

..__._

,

m ~

I

i

---

Jl

,'l !%.

-....,~j i ~ •

^

i

Y

w

._...

Double hehces

Single hehces

(

) (b)

FIG. 9. Models of the two classes of Vinca-induced paracrystals. (a) Single helices. Helices labled A are 180° out of phase with those marked B. Contacts are found formed between A and B helices but not between A and A or B and B. (b) Double helices. The two helices in each column are 180° out of phase with each other. Each column makes a contact with all neighboring columns. Reprinted from Amos et al. (1984), with permission of the authors and the copyright holder, Academic Press, London. ~T

51/2--H

266

R.H. HIMES

tion electron microscopy has been used to show a difference in structure between the sea urchin egg paracrystals produced in vivo and the brain paracrystals produced in vitro (Amos et al., 1984). The paracrystals are composed of macrotubules packed in parallel along their long axis. The macrotubules in brain paracrystals consist of a single protofilament as a helix while those in the sea urchin egg paracrystals are composed of a pair of helices 180 ° out of phase (Fig. 9). The requirement for a high M g 2+ concentration to produce the packed helical arrays in vitro is consistent with the proposal that either MAPs, Mg 2+ or other cationic species are required to produce such structures (Haskins et al., 1981).

7. C O N C L U S I O N S The dimeric Vinca alkaloids bind to one intrinsic site in the tubulin dimer. This site is linked to a protein self-association reaction which has made it impossible to study the binding reaction in the absence of tubulin self-association. Solution variables, especially Mg 2+ , have a dramatic effect on the selfassociation constant and perhaps on the intrinsic binding constant. The great variation in reported apparent binding constants is due to this effect. The limited number of binding studies using synthetic derivatives of VLB and V C R show relatively small changes in apparent binding constants accompanying changes of substituents at the N - l , C-3, C-4 and C-4' positions. However, in vitro assembly assays demonstrated the essential nature of the methoxy carbonyl at C-18' and the correct configuration at C-18' and C-T. The Vinca alkaloids also bind to microtubules with apparent high affinity at the microtubule ends which inhibits tubulin dimer addition, and low affinity along the walls of the microtubules which results in the separation of the protofilaments and the formation of protofilaments coiled into spirals. These structures apparently associate to form paracrystalline structures. Acknowledgements--I wish to thank Drs Serge N. Timasheff, Mary Ann Jordan, Leslie Wilson, and William D. Singer for their critical reading of the manuscript and helpful suggestions.

REFERENCES AMOS, L. A., JUBB, J. S., HENDERSON, R. and VIGERS,G. (1984) Arrangement of protofilaments in two forms of tubulin crystal induced by vinblastine. J. molec. Biol. 178: 711-729. BENSCH,K. G. and MALAWISTA,S. E. (1969) Microtubular crystals in mammalian cells. J. Cell Biol. 40: 95-107. BENSCH,K. G., MARANTZ,R., WISNIEWSKI,H. and SHELANSKI, M. (1969) Induction in vitro of microtubular crystals by Vinca alkaloids. Science 165: 495-496. BHATTACHARYYA,B. and WOLFE,J. (1976) Tubulin aggregation and disaggregation: mediation by two distinct vin-

blastine-binding sites. Proc. Bath. Acad. Sci. U.S.A. 73: 2375-2378. BORMAN,U S. and KUEHNE,M. E. (1989) Specific alterations in the biological activities of C-20'-modified vinblastine congeners. Biochem. Pharmac. 38: 715-724. BRYAN'. J. (1971) Vinblastine . t~ and microtubules I Induction and isolation of crystals from sea urchin oocytes. Expl. Cell Res. 66: 129-136. BRYAN,J. (1972a) Vinblastine and microtubules. II. Characterization of two protein subunits from the isolated crystals. J. molec. Biol. 66:157 168. BRYAN,J. (1972b) Definition of three classes of binding sites in isolated microtubule crystals. Biochemistry 11: 2611-2616. CONRAD, R. A., CULLINAN,G. J., GERZON,K. and POOKE, G. A. (1979) Structure-activity relationships of dimeric Catharanthus alkaloids. 2. Experimental antitumor activities of N-substituted deacetylvinblastine amide (vindesine sulfates). J. reed. Chem. 22: 391-400. CUTTS, J. n . , BEER, C. T. and NOBLE, R. L. (1960) Biological properties of vincaleukoblastine, an alkaloid in Vinea rosea Linn, with reference to its antitumor action. Cancer Res. 20: 1023-1031. DONOSO, J. A., HASKINS, K. M. and HIMES, R. H. (1979) Effect of microtubule-associated proteins on the interaction of vincristine with microtubules and tubulin. Cancer Res. 39: 1604-1610. ERICKSON,H. P. (1975) Negatively stained vinblastine aggregates. Ann. N.Y. Acad. Sci. 253: 51-52. FRIGON, R. P. and TIMASHEFF, S. N. (1975a) Magnesiuminduced self-association of calf brain tubulin. I. Stoichiometry. Biochemistry 14: 4559-4566. FR1GON, R. P. and TIMASrtEFF,S. N. (1975b) Magnesiuminduced self-association of calf brain tubulin. II. Thermodynamics. Biochemistry 14: 4567-4573. FUJIWARA,K. and TILNEY,L. G. (1975) Substructural analysis of the microtubule and its polymorphic forms. Ann. N.Y. Acad. Sci. 253: 27-50. GEORGE, P., JOURNEY, L. J. and GOLDSTEIN,M. N. (1965) Effects of vincristine on the fine structure of HeLa cells during mitosis. J. Bath. Cancer Inst. 35:355 361. GERZON, K. (1980) Dimeric Catharanthus alkaloids. In: Antieancer Agents Based on Natural Product Models,

pp. 271-316, CASSADY, J. M. and DOUROUS,J. D. (eds) Academic Press, New York. HASK1NS, K. M., DONOSO, J. A. and HIMES, R. H. (1981) Spirals and paracrystals induced by vinca alkaloids: evidence that microtubule-associated proteins act as polycations. J. Cell. Sci. 47: 237-247. JOHNSON,I. S., WRIGHT,H. F., SVOBOOA,G. H. and VLANTIS, J. (1960) Antitumor principles derived from Vinca rosea Linn. I. Vincaleukoblastine and leurosine. Cancer Res. 20: 1016-1022. JOHNSON, I. S., ARMSTRONG, J. G., GORMAN, M. and BURNETT, J. P. (1963) The Vinca alkaloids: a new class of oncolytic agents. Cancer Res. 23: 1390-1427. JORDAN, M. A. and WILSON, L. (1990) Kinetic analysis of tubulin exchange at microtubule ends at low vinblastine concentrations. Biochemistry 29: 2730-2739. JORDAN, M. A., HIMES, R. H. and WILSON,L. (1985) Comparison of the effects of vinblastine, vincristine, vindesine, and vinepidine on microtubule dynamics and cell proliferation in vitro. Cancer Res. 45:2741 2747. JORDAN,M. A., MARGOLIS,R. L., HIMES,R. H. and WILSON, L. (1986) Identification of a distinct class of vinblastine binding sites on microtubules. J. molec. Biol. 187: 61-73. JORDAN, M. A., THROWER, D. and WILSON, L. (1991) Mechanism of inhibition of cell proliferation by vinca alkaloids. Cancer Res. 51:2212 2222. KRISI-IAN,A. (I 968) Time-lapse and ultrastructure studies on the reversal of mitotic arrest induced by vinblastine sulfate in Earle's L cells. J. Bath. Cancer Inst. 41: 581-586.

Interactions of the catharanthus alkaloids LEE, J. C., HARRISON, D. and T1MASHEFF,S. N. (1975) Interaction of vinblastine with calf brain microtubule protein. J. biol. Chem. 250: 9276-9282. LUDUEfiA, R. F., FELLOUS, A., FRANCON, J., NtmEZ, J. and MCMANUS, L. (1981) Effect of tau on the vinblastineinduced aggregation of tubulin. J. Cell Biol. 89: 680~83. LUDUEI~IA,R. F., FELLOUS,A., MCMANUS, L., JORDAN,M. A. and NUNEZ, J. (1984) Contrasting roles of tau and microtubule-associated protein 2 in the vinblastine-induced aggregation of brain tubulin. J. biol. Chem. 20: 12890-12898. MARANTZ, R., VENTILLA,M. and SHELANSKI,M. (1969) Vinblastine-induced precipitation of microtubule protein. Science 165: 498~499. NA, G. C. and TIMASHE~, S. N. (1980a) Stoichiometry of the vinblastine-induced self-association of calf brain tubulin. Biochemistry 19: 1347-1354. NA, G. C. and TIMASHEFF, S. N. (1980b) Thermodynamic linkage between tubulin self-association and the binding of vinblastine. Biochemistry 19: 1355-1365. NA, G. C. and TIMASHEFF,S. N. (1982) In vitro vinblastineinduced tubulin paracrystals. J. biol. Chem. 257: 10387-10391. NA, G. C. and TIMASHEFF, S. N. (1986a) Interaction of vinblastine with calf brain tubulin: multiple equilibria. Biochemistry 25: 6214-6222. NA, G. C. and TIMASrm~ S. N. (1986b) Interaction of vinblastine with calf brain tubulin: effects of magnesium ions. Biochemistry 25: 6222~228. NASlOULAS, G., GRAMMBITTER,K., HIMES, R. H. and PONSTINGL, H. (1990) Interaction of a new photosensitive derivative of vinblastine, NAPAVIN, with tubulin and microtubules in vitro. Eur. J. Biochem. 192: 69-74. NOBLE, R. L., BEER, C. T. and CUTTS, J. H. (1958a) Role of chance observations in chemotherapy: Vinca rosea. Ann. N.Y. Acad. Sci. 76: 882-894. NOBLE, R. L., BEER, C. T. and CUTTS, J. H. (1958b) Further biological activities of vincaleukoblastine--an alkaloid isolated from Vinca rosea (L). Biochem. Pharmac. I: 347-348. OLMSTED,J. B., CARLSON, K., KLEBE, R., RUDDLE, F. and ROSENBAUM, J. (1970) Isolation of microtubule protein from cultured mouse neuroblastoma cells. Proc. natn. Acad. Sci. U.S.A. 65: 129-136. OWELLEN, R. J., OWENS, A. H. JR. and DONIGAN, D. W. (1972) The binding of vincristine, vinblastine and colchicine to tubulin. Biochem. biophys. Res. Commun. 47: 685~59 I. OWELLEN, R. J., HARTKE, C. A., DlCKERSON, R. M. and HAINS, F. O. (1976) Inhibition of tubulin-microtubule polymerization by drugs of the Vinca alkaloid class. Cancer Res. 36: 1499-1502. OWELLEN,R. J., DONIGAN,D. W., HARTKE,C. A. and HAINS, F. O. (1977) Correlation of biological data with physicochemical properties among the Vinca alkaloids and their congeners. Biochem. Pharmac. 26: 1213-1219. POTIER, P., GUI~NARD, D. and ZAVALA,F. (1979) R6sultats r6cents dans le domaine des alcaloi'des antitumoraux du groupe de la vinblastine. Etudes biochimiques. C.R. Soc. Biol. 173: 414-424.

267

PRAKASH, V. and TIMASHEFF,S. N. (1983) The interaction of vincristine with calf brain tubulin. J. biol. Chem. 258: 1689-1697. PRAKASH, V. and TIMASrmFF, S. N. (1985) Vincristine-induced self-association of calf brain tubulin. Biochemistry 24: 5004-5010. PRAKASH, V. and T I M A S ~ , S. N. (1991) Mechanism of interactionof Vinca alkaloidswith tubulin:catharanthine and vindoline. Biochemistry 30: 873-880. SAFA, A. R. and FELSTED, R. L. (1987) Specific Vinca alkaMid-binding polypeptides identifiedin calf brain by photoaliinitylabeling.J. biol. Chem. 262: 1261-1267. SAFA, A. R., HAMEL, E. and FELSTED, R. L. (1987) Photoaffinitylabeling of tubulin subunits with a photoactive analogue of vinblastine.Biochemistry 26: 97-102. SCHOCHET, S. S. JR., LAMPERT, P. W. and EARLE, K. M. (1968) Neuronal changes induced by intrathecal vincristine sulfate. J. Neuropath. exp. Neurol. 27: 645~58. SINGER, W. D., HERSH, R. T. and HI,mS, R. H. (1988) Effect of solution variables on the binding of vinblastine to tubulin. Biochem. Pharmac. 37: 2691-2696. SINGER, W. D., JORDAN, M. A., WILSON, L. and HIMES, R. H. (1989) Binding of vinblastine to stabilized microtubules. Molec. Pharmac. 36: 366-370. STEBBINGS, H. (1971) Influence of vinblastine sulphate on the deployment of microtubules and ribosomes in telotrophic ovarioles. J. Cell. Sci. 8: I 11-125. SVOBODA,G. n . (1961) Alkaloids of Vinca rosea (Catharanthus roseus). IX. Extraction and characterization of leurosidine and leurocristine. Lloydia 24: 173-178. WARFIELD,R. K. N. and BOUCK, G. B. (1974) Microtubulemacrotubule transitions: intermediates after exposure to the mitotic inhibitor vinblastine. Science 186: 1219-1221. WEISENBERG,R. C. and TIMASHEFF,S. N. (1970) Aggregation of microtubule subunit protein. Effects of divalent cations, colchicine and vinblastine. Biochemistry 9: 4110-4116. WILSON, L. (1970) Properties of colchicine binding protein from chick embryo brain. Interactions with Vinca alkaloids and podophyllotoxin. Biochemistry 9: 4999-5007. WILSON, L. and FRIEDKIN, M. (1967) The biochemical events of mitosis. II. The in vivo and in vitro binding of colchicine in grasshopper embryos and its possible relation to inhibition of mitosis. Biochemistry 6: 3126-3135. WILSON, L., BRYAN, J., RUBY, A. and MAZIA, O. (1970) Precipitation of proteins by vinblastine and calcium ions. Proc. natn. Acad. Sci. U.S.A. 66: 807-814. WILSON, L., CRESWELL,K. M. and CHIN, D. (1975) The mechanism of action of vinblastine. Binding of [acetyl3H]vinblastine to embryonic chick brain tubulin and tubulin from sea urchin sperm tail outer doublet microtubules. Biochemistry 14: 5586-5592. W1LSON, L., MORSE, A. N. C. and BRYAN, J. (1978) Characterization of acetyl-3H-labeled vinblastine binding to vinblastine-tubulin crystals. J. molec. Biol. 121: 255-268. WILSON, L., JORDAN, M. A., MORSE,A. and MARGOLIS,R. L. (1982) Interaction of vinblastine with steady-state microtubules in vitro. J. molec. Biol. 159: 125-149. ZAVALA,F., Gt~NARD, D. and POTIER, P. (1978) Interaction of vinblastine analogues with tubulin. Experientia 34: 1497-1499.