Proc. Natl. Acad. Sci. USAVol. 74, No. 8, pp. 3466-3470, August 1977Cell Biology

Addition of colchicine-tubulin complex to microtubule ends:The mechanism of substoichiometric colchicine poisoning

(protein self-assembly/drug binding)

ROBERT L. MARGOLIS AND LESLIE WILSONDepartment of Biological Sciences, University of California, Santa Barbara, Santa Barbara, California 93106

Communicated by Daniel Mazia, June 8, 1977

ABSTRACT Colchicine blocks microtubule polymerizationby an unusual substoichiometric poisoning mechanism. Wehave investigated the mechanism by which this poisoning oc-curs with several experimental approaches,-and have found thatcolchicine acts by addition to microtubule ends as a colchi-cine-tubulin complex.

Colchicine, a potent drug that interferes with microtubule as-sembly both in vivo and in vitro, binds to the dimeric subunitof the microtubule, tubulin, with a stoichiometry of 1 mol/mol of dimer (1-6). In assembled microtubules most drugbinding sites are blocked (ref 5; W. Rodgers, R. Margolis, andL. Wilson, unpublished data). However, it has been impossibleto rule out the binding of colchicine to a limited number of sites,for instance, those uniquely exposed on either of the two mi-crotubule ends.The mechanism by which colchicine poisons microtubule

assembly has remained to be elucidated. Olmsted and Borisy(4) showed that amounts of colchicine sufficient to bind to onlya small percentage of soluble dimers could effectively poisonmicrotubule assembly in vitro. It may also be inferred that asimilar poisoning event occurs during exposure in living cells.Taylor has estimated mitotic blockage occurs when only 3-5%of the tubulin is complexed with colchicine in KB cells (2).

There are two attractive mechanisms that might account forthe substoichiometric poisoning of microtubule assembly withcolchicine. Tubulin molecules on the microtubule free ends mayhave a much higher affinity for the drug than does soluble tu-bulin, thereby permitting a preferential binding of colchicineto the microtubule ends; or colchicine may bind first to a limitednumber of soluble dimers, then these colchicine dimer com-plexes (CD complexes) add to the microtubule ends duringassembly, thus blocking further dimer addition.We report here that the latter mechanism appears correct.

Colchicine poisons microtubule assembly by first binding to thesoluble 6S dimer, which then adds to the growing microtubuleas a colchicine-dimer camplex during the normal process ofassembly. Once added, the CD complex effectively "caps" themicrotubule and aborts further polymerization. Of the freedimers in solution, only a small percentage need be complexedwith colchicine for assembly to be blocked.

MATERIALS AND METHODSTubulin Preparation. Beef brain tubulin was prepared by

three cycles of polymerization-depolymerization using amodification (C. F. Asnes and L. Wilson, to be published) of theBorisy et al. procedure for porcine brain (7), in an assembly

The costs of publication of this article were defrayed in part by thepayment of page charges from funds made available to support theresearch which is the subject of the article. This article must thereforebe hereby marked "advertisement" in accordance with 18 U. S. C.§1734 solely to indicate this fact.

buffer composed of 20mM sodium phosphate, 100mM sodiumglutamate, 2.5 mM GTP, 1 mM MgCl2, and 0.5 mM [eth-ylenebis(oxyethylenenitrilo)]tetraacetate (EGTA) at pH 6.75.All experiments with beef brain tubulin were carried out in thisbuffer unless indicated otherwise. The beef brain microtubuleprotein preparation contained approximately 75% tubulin, and25% high-molecular-weight microtubule-associated proteins.

Chick brain tubulin was extracted at 00 from 13- to 16-dayembryos by Dounce homogenization in 1 ml per brain of buffercomposed of 100 mM piperazine-N,N'-bis(2-ethanesulfonicacid) (Pipes) (Calbiochem.), 1.0mM MgCl2, and 10mM EGTA,pH 6.9. After centrifugation at 30,000 X g for 10 min at 20,GTP (final concentration 2.5 mM) was added to the superna-tant, which was centrifuged at 2° at either 30,000 X g for 20min, or at 200,000 X g for 60 min. Resulting supernatants wereused for the electron microscopy (EM) experiments. All ex-periments with chick brain microtubule protein were carriedout with the above buffer.

Total protein was determined by the method of Lowry et al.(8), and tubulin concentration was determined by measurementof areas under curves from 640-nm scans of fast-green-stainedsodium dodecyl sulfate/polyacrylamide gels (9).

Electron Microscope Assay of Assembly Blockage. Chickembryo brain tubulin contained in a 30,000 X g supernatant(total protein 10 mg/ml) was polymerized for 4 min at 300 toobtain microtubule fragments and then placed on a 0.5%Formvar/carbon-coated EM grid. Those fragments that ad-hered to the grid in 20 sec were used as seeds for further as-sembly. Grids were rinsed with chick brain buffer and dippedfor 10 sec in buffer solution containing DEAE-dextran at 0.7mg/ml (Sigma) (10) then rinsed with buffer at 220. Grids werethen everted onto a 200,000 X g supernatant containing 2.5mMGTP (total protein 6 mg/ml) and incubated an additional 4 minat 220. Individual colchicine incubations were performed asdescribed in Results, at a final concentration of 100,uM. Gridswere then negatively stained with 1% uranyl acetate accordingthe procedure of Olmsted et al. (10), and observed with aPhilips 300 electron microscope.

Light Scattering Assay of Assembly Blockage. Preparationof CD Complex. CD complex was formed by incubation ofthree-times-recycled beef brain microtubule protein (totalprotein 1 mg/ml, in 20mM sodium phosphate/100mM sodiumglutamate, pH 6.75) with 100 ,gM colchicine for 30 min at 300.The solution was chilled to 00 and CD complex and unboundtubulin were separated from free colchicine by gel filtrationon a 1 X 18-cm column of Bio-Gel P-10 (Bio-Rad Laboratories).The quantity of CD complex used in these experiments wasascertained by inclusion of [3H]colchicine (New England Nu-

Abbreviations: CD complex, tubulin dimer containing bound colchi-cine; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetate; EM, elec-tron microscopy.

3466

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

r.

Proc. Natl. Acad. Sci. USA 74 (1977) 3467

A I

;'1.:

C

he ;.._.*;anAn..

<_dS +FIG. 1. Chick embryo brain 30,000 X g supernatant was incubated for 4 min at 300 to form microtubules. These short microtubules were affixed

to EM grids, exposed to colchicine as described in the text, decorated with DEAE-dextran, and tested for colchicine blockage by exposure to200,000 x g superhatants containing 6S tubulin under assembly conditions. Chick embryo brain microtubules do not depolymerize upon drugexposure in vitro (ref. 12; K. Anderson and L. Wilson, unpublished data). Similarly, the grid-bound microtubule fragments that we observedappeared, by EM, to be stable to all our experimental manipulations. (A) Two blocked fragments with double wall decorations, after exposureto colchicine in the presence of tubulin, are shown. Bar = 0.2 ,m. (B) Higher magnification of a fragment blocked by CD complex. Decorations,with this technique, were uniform and highly visible. Bar = 0.1 jsm. (C) A control microtubule with a long undecorated tail (single wall) emergingfrom an end decorated region (arrow). Bar = 0.2 Mm.

clear) (6). Undt-r the incubation conditions used, approximately50% of the tubulin isolated from the column was complexedwith colchicine. The half-time for dissociation of the complexwas no less thall 22 hr under the conditions employed. Thus,no more than 1.5% of the complex could have dissociated duringthe course of these experiments.

Quantitation of the Effect ofCD Complex on MicrotubulePolymerization. Microtubule assembly at 300 was followedwith use of light scattering at 350 nm or by viscometry (4, 11).For light scattering measurements a Gilford recording spec-trophotometer (model 2400) with a constant-temperature cu-vette chamber was used. Whenever colchicine was presentduring incubation in high concentration, measurements weremade by viscometry, because the drug absorbs strongly at 350nm. Total protein concentrations in these experiments were 1mg/ml (2 mg/ml for viscometry).

RESULTSEither colchicine binds directly to the microtubule end, or thedrug binds first to a dimer, which then adds to the microtubuleend and "caps" it. The following experiments were designedto distinguish between these two possibilities.

Electroh Microscopy Visualization of Poisoning by CDComplex. Tubulin contained in a chick embryo brain 30,000X g supernatant was polymerized to form short microtubulefragments, which were then placed on an EM grid. The ad-hering fragments were coated by a 10-sec exposure to DEAE-

dextran. Further incubation with tubulin created a "decoration"(10) which distinctly marked that portion of the microtubulethat had been exposed to DEAE-dextran. Thus, the decorationserved as a marker for the microtubule's competency to as-semble before and after DEAE-dextran exposure. The grid wasthen exiosed to tubulin in a 200,000 X g supernatant, whichpolymerized only onto preformed microtubule fragments (10).The typical result for such a control was a population of mi-crotubules, most of which had short (0.25 i 0.02 Am) decoratedregions and long (1.72 ± 0.28 ,um) undecorated "tails" (Fig. 1C).Decorated regions formed by this grid incubation techniquewere uniformly distinct and uninterrupted.To determine if colchicine adds to microtubule fragments

and prevents their further polymerization, we exposed mi-crotubule fragments on grids, rinsed free of dimers, to 100 AMcolchicine (sufficient to totally block assembly), decorated them,then challenged their polymerization-seeding capacity by in-cubation for 4 min at 220 with 200,000 X g supernatant protein.Alternatively, to determine if free dimers are a necessary in-termediate in colchicine addition to microtubules, short mi-crotubule segments were made, as above, and colchicine (100AM) was added to the dimer microtubule mixture for an ad-ditional 2-min (30°) incubation. Microtubule fragments thatresulted Were then allowed to adhere to grids and assayed asabove for additions to decorated regions.The results were quantified by scanning grids and scoring

every decorated region seen in a random scan for presence or

Cell Biology: Matgolis and Wilson

3

-Jr-

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

3468 Cell Biology: Margolis and Wilson

.0.10 -0 'I

0.05 / .,

0-

8 16 24 32 40Min

FIG. 2. The effect of added CD complex on microtubule assemblywas assayed by turbidity measurements of recycled beef brain tubulinpolymerization at 300 (see Materials and Methods). Relative tocontrol assembly ( ), the assembly rate of tubulin containing 2.6%CD complex (---) was inhibited 41%. When CD complex (2.6% of thezero-time free tubulin concentration) was added at 16 min (- -),immediate blockage occurred. The control contained total proteinat 1 mg/ml, tubulin at 0.75 mg/ml. CD complex was added in 100 JI,containing protein at 0.5 mg/ml, to a total final volume of 0.85 ml.Therefore, dilution of the total incubation volume was only 12% anddilution of total protein concentration was 6%.

absence of an undecorated tail at least the length of the deco-rated region. We found that 19.7% of microtubules on controlgrids did not elongate. If microtubules were exposed to col-chicine in the absence of dimers, 15.4% did not elongate.Therefore, the presence of colchicine, by itself, was not suffi-cient to block assembly. If microtubules were exposed to thedrug in the presence of dimers, 65.2% of the microtubules wereblocked from further assembly. Fig. 1 compares decoratedfragments blocked by CD complex (Fig. 1 A and B) with atypical augmented microtubule (no colchicine, Fig. IC).

Because microtubules or microtubule-dimer mixtures wereexposed to colchicine for only 2 min in the EM experiment, wetested the binding efficiency of 100 ,gM colchicine to tubulinin this time period, using a Bio-Gel P-10 column to separate[3H]colchicine-CD complex from free [3H]colchicine. Ninepercent of the dimers in solution had bound the drug in thistime.CD Complex Poisoning of Assembly. If dimers mediate

colchicine action in poisoning microtubule assembly, CDcomplex in the absence of free colchicine should exhibit thecapacity to poison assembly. CD complex, when isolated fromfree colchicine on a Bio-Gel P-10 column, showed the expectedbiochemical activity. Microtubule assembly, as assayed by lightscattering, had a lesser rate and extent of polymerization whenapproximately 1 of 40 dimers in solution was complexed tocolchicine (Fig. 2). In replicate experiments, a CD complex tofree tubulin ratio of 0.026 (10.001) yielded 0.41 (+0.02) of thecontrol polymerization rate. The poisoning effect was morepronounced if CD complex was added later during polymer-ization (Fig. 2), possibly because the ratio of CD complex to freetubulin was much higher in late stages of polymerization.

Addition of CD Complex to Preformed Microtubules.When CD complex was added to a solution of beef brain tubulinafter polymerization had reached a plateau, there was nomeasurable depolymerization. In fact, all that was observed wasa limited and transient dilutiqn effect upon addition of CDcomplex (Fig. 2). Viscometry measurements of the effect of freecolchicine addition to tubulin after polymerization was com-plete showed a similar resistance to disassembly (Fig. 3). Aconcentration of colchicinre' sufficient to totally block assembly(100 ,uM) produced only limited depolymerization of mi-

0.80

0.60

0.40

0.20

0a

,,0

, , 0

00

0

10 20 30 40 50Min

FIG. 3. Specific viscosity (%p) measurements were made of thebeef brain microtubule response to colchicine in vitro. When colchi-cine (100 ,sM) was added at the start of 300 incubation (@-@), noassembly was evident. Two other samples, assembled to plateau beforecolchicine (100 AM) was added at 18 min (---A), or at 42 min(0- - -0), showed limited disassembly. In these experiments, 20,l of colchicine solution was added to a sample volume of 2.0 ml.

crotubules once formed. EM samples of this material showedintact microtubules still present after prolonged incubation withdrug. The quantity of microtubules could not be visually dis-tinguished from controls. Recycled beef brain microtubuleswere similarly stable to podophyllotoxin (50,uM), vinblastine(100 AM), and GDP (1 mM), all at concentrations sufficient tototally block assembly.

Potentiation of Poisoning by Preincubation. Colchicine,at low concentrations, binds slowly to tubulin (6). Underpolymerization conditions, the drug must compete with theassembly reaction for free dimers. Microtubules are relativelyinsensitive to colchicine-induced depolymerization in vitro,thus, assembly permanently removes dimers from access tocolchicine. Therefore, if the CD complex mediates colchicinepoisoning, one will obtain an inaccurately low inhibition con-stant, and preincubation of tubulin with a low concentrationof colchicine, before assembly is allowed to occur, should po-tentiate the effect of the drug. Beef brain microtubule protein(protein at 1 mg/ml, 6.8 MM tubulin) was polymerized in thepresence of colchicine and the assembly was followed with lightscattering. Results (Fig. 4A) where there was no preincubation

A 0.06 BNo Pre-incubation Pre-incubation

0.15

0.0451 0.10

0

005 ~~~~~~~~0.02

0 0 ---

0 4 8 12 16 20 0 4 8 12 16 20Min

FIG. 4. The substoichiometric poisoning effect of colchicine onrecycled beef brain tubulin assembly was assayed by turbidity mea-surements at 300. (A) The protein was not preincubated with col-chicine before the assembly reaction. A limited substoichiometricpoisoning effect of colchicine on assembly was seen. (B) If the tubulinwas preincubated for 30 min at 300 with colchicine before assemblywas initiated by the addition of GTP, the poisoning effect was sub-stantially potentiated. Total protein concentration was 1 mg/ml.Control; ---, 1 gM colchicine;- - -, 0.2 sM colchicine.

Proc. Natl. Acad. Sci. USA 74 (1977)

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

Proc. Natl. Acad. Sci. USA 74 (1977) 3469

showed an 8% inhibition at 0.20AM and 23% inhibition at 1.0MM colchicine. Preincubation with drug for 30 mim fat 3086,before GTP was added and polymerization proceeded, yielded(Fig. 4B) 60% inhibition at 0.20AM and 100% inhibition at 1.0MM colchicine (measured relative to a preincubated control).The molar ratio (0.034) of colchicine to tubulin [CD complexto tubulin ratio of 0.023 as determined by dextran-charcoalcolchicine binding assay (6)] that yielded 60% inhibition ofassembly following preincubation was in close agreement withthe molar ratio of CD complex to tubulin (0.026) that yielded60% inhibition in the CD complex poisoning experiment (Fig.2). In that experiment, we determined the half time of disso-ciation of cokhicine from tubulin to be 1400 min at 00, and thusestimate no more than 1.5% of the CD complex reverted to freedimer and drug in the 30 min it took to isolate CD complex andbegin kinetic measurements. We therefore believe the CDcomplex to tubulin ratios derived from the two different ex-perimental approaches are reasonably accurate, and that theclose agreement between them is real.

DISCUSSIONThree independent experimental results each indicate thatcolchicine inhibits microtubule assembly by means of a CDcomplex. Colchicine first binds to free dimers, and the infre-quent dimers that bind the drug add to the microtubule endsduring the course of assembly. These CD complexes then capthe microtubules, making further assembly impossible. Thesubstoichiometric blockage of microtubule assembly (4) bycolchicine is explained by the mechanism of poisoning revealedin these studies.

Other drugs that interfere with microtubule assembly, vin-blastine and podophyllotoxin, also poison polymerization sub-stoichiometrically (12). The possibility exists that these drugspoison microtubule assembly through a similar mechanism.

It is reported that microtubules are in a steady-state conditionduring assembly in vitro, depolymerization occurring simul-taneously with polymerization (13). If such a steady-statecondition exists, microtubule poisoning by an infrequent CDaddition necessitates that disassembly not occur at the site ofCD complex addition. If 1 of every 40 dimers contained col-chicine, each CD complex released from a microtubule wouldallow an average of 39 dimers to add before blockage againoccurred. The net effect, a poisoning of assembly, would benullified.One or more of three possibilities can explain the success of

the in vitro poisoning phenomenon: (i) the CD complex mayhave an increased affinity for the microtubule end as comparedwith an uncomplexed dimer, making dissociation of the cappingcomplex unusually slow or perhaps nonexistent; (ii) depoly-merization may not normally occur at the site of CD complexaddition; or (iii) a steady-state may not exist under our assemblyconditions in vitro.A conformational shift that might block depolymerization

seems plausible. The existence of a true "cap" that freezes themicrotubule end is an attractive concept, although it is seem-ingly contradicted by evidence that colchicine causes somedepolymerization in vitro until a new "steady state" is attained(4) (see Fig. 3).One alternate possibility is that disassembly is not occurring

at the site of assembly. Microtubule assembly occurs as a biasedpolar phenomenon (10, 14). This means that one end of themicrotubule (designated the X end) augments to a much greaterextent per unit time than does the other (Y end). Suppose thatthe end opposite the primary addition site is the primary de-polymerization site (Y). Then a colchicine poisoned population

of- microtuibules will reach a new steady state determined byeuIfibrum conditions at the primary site of depolymerization(Y). If the CD complex is sterically unable to add to the mi-crotubule Y end, and the X end is CD capped, a new equilib-rium will be determined by concentrations of free tubulin andY ends, independent of the CD complex population. The Y endis capable of net assembly (10, 14) and conceivably could reacha new steady state in the presence of CD complex, yielding anew microtubule population plateau.The third possibility for the mechanism of substoichiometric

CD complex poisoning of assembly, that a true equilibrium isnot present in our in vitro microtubule system, also remainsviable.

Perhaps, the response of the microtubule population in vitroto drug reflects a combination of preferential Y end depoly-merization and regional blockage to depolymerization alongthe microtubule length. It has been reported that microtu-bule-associated proteins interfere with drug-induced depoly-merization in vitro (15). Drug-poisoned microtubules possibly,therefore, depolymerize until a region of high concentrationof microtubule-associated protein is reached.The situation within the cell is evidently different. Cyto-

plasmic and mitotic apparatus microtubules are quite labile tocolchicine and other drugs. It is reasonable to assume that col-chicine and other drugs act in the cell as they do in vitro, bypoisoning assembly substoichiometrically. Podophyllotoxin, adrug that resembles colchicine and shares a binding site ontubulin with colchicine, poisons microtubule assembly in vivo.Knowing the concentration of free tubulin in a sea urchin em-bryo (25 ,M) (16), and the binding constant for podophyllotoxinat 130 (0.76 MAM), one can estimate the ratio of podophyllo-toxin-dimer complex to free dimer present when mitosis isinhibited by 50% (at a podophyllotoxin concentration of 0.096uM). Less than 1 of 250 dimers need be drug bound for mitoticblockage to occur (C. Rauch and L. Wilson, unpublished ob-servations). This low ratio is similar to that required to block invitro assembly of brain microtubules (12).

Microtubules extensively depolymerize in cells when poi-soned substoichiometrically with drug. This phenomenonpresents most clearly the paradox outlined above, that disas-sembly cannot occur if it first releases the drug-dimer complex,because this releases the microtubule for further assembly.Poisoning can only occur if the drug caps the microtubule endand alters its ability to disassemble, or if disassembly prefer-entially occurs at the end opposite the assembling end underphysiological conditions.

These results strongly suggest that microtubule disassemblyfollowing exposure of cells to drug occurs only when each mi-crotubule is in active equilibrium, both assembling and disas-sembling in the absence of drug. Thus differences in microtu-bule sensitivity to mitostatic poisons (17) may reflect in partdifferent intrinsic equilibria in different microtubule popula-tions.

We thank W. Rodgers for excellent technical assistance. This workwas supported by American Cancer Society Grant CH4B, U.S. PublicHealth Service Grant NS13560, and Anna Fuller Fund PostdoctoralFellowship 443 to R.L.M. A preliminary report on this work was pre-sented at the 61st Annual Meeting of the Federation of American So-cieties for Experimental Biology (18).

1. Inoue, S. (1952) Exp. Cell Res. Suppl. 2, 305-314.2. Taylor, E. (1965) J. Cell Biol. 25, 145-160.3. Borisy, G. G. & Taylor, E. (1967) J. Cell Biol. 34, 525-533.4. Olmsted, J. B. & Borisy, G. G. (1973) Biochemistry 12, 4282-

4289.

Cell Biology: Margolis and Wilson

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021

3470 Cell Biology: Margolis and Wilson

5. Wilson, L. & Meza, I. (1973) J. Cell Biol. 58,709-719.6. Wilson, L. & Bryan, J. (1974) Adv. Cell Mol. Biol. 3,21-71.7. Borisy, G. G., Marcum, J. M., Olmsted, J. B., Murphy, D. 13. &

Johnson, IK. A. (1975) Ann. N.Y. Acad. Sd. 253,107-132.8. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J.

(1951) J. Biol. Chem. 193,265-275.9. Weber, K. & Osborn, M. (1969) J. Biol. Chem. 244, 4406-

4413.10. Olmsted, J. B., Marcum, J. M., Johnson, K. A., Allen, C. & Borisy,

G. G. (1974) J. Supramole. Struct. 2,429-450.11. Gaskin, F., Cantor, C. R. & Shelanski, M. L. (1975) Ann. N.Y.

Acad. Sci. 253,133-145.

Proc. Nati. Acad. Sci. USA 74 (1977)

12. Wilson, L., Anderson, K. & Chin, D. (1976) in Cell Motility (ColdSpring Harbor Conferences on Cell Proliferation, Cold SpringHarbor, N.Y.), Vol. 3. pp. 1051-1064.

13. Johnson, K. A. & Borisy, G. G. (1974) Fed. Proc. 33, 1231.14. Dentler, W., Granett, S., Witman, G. & Rosenbaum, J. (1974)

Proc. Natl. Acad. Sci. USA 71,1710-1714.15. Haga, T. & Kurokawa, M. (1975) Biochim. Biophys. Acta 392,

335-345.16. Pfeffer, T. A., Asnes, C. & Wilson, L. (1976) J. Cell Biol. 69,

599-607.17. Behnke, 0. & Forer, A. (1967) J. Cell Sci. 2, 169-192.18. Margolis, R. L. & Wilson, L. (1977) Fed. Proc. 36,899.

Dow

nloa

ded

by g

uest

on

Janu

ary

2, 2

021