Podophyllotoxin as a Probe for the Colchicine Binding Site of Tubulin

The binding of [YH]podophyllotoxin to tubulin, measured by a DEAE-cellulose filter paper method, occurs with an affinity constant of 1.8 x 10” M-’ (37” at pH 6.7). Like colchicine, -0.8 mol of podophyllotoxin are bound per mol of tubulin dimer, and the reaction is entropy-driven (43 cal deg-* mol-I). At 37” the association rate constant for podophyllotoxin binding is 3.8 x 10fi M-' h-l, -10 times higher than for colchicine; this is reflected in the activation energies for binding which are 14.7 kcal/mol for podophyllotoxin and 20.3 kcal/mol for colchicine. The dissociation rate constant for the tubulin-podophyllotoxin complex is 1.9 h-l, and the affinity constant calculated from the ratio of the rates is close to that obtained by equilibrium measurements. and mutually competitive

The binding of [YH]podophyllotoxin to tubulin, measured by a DEAE-cellulose filter paper method, occurs with an affinity constant of 1.8 x 10" M-' (37" at pH 6.7). Like colchicine, -0.8 mol of podophyllotoxin are bound per mol of tubulin dimer, and the reaction is entropy-driven (43 cal deg-* mol-I).
At 37" the association rate constant for podophyllotoxin binding is 3.8 x 10fi M-' h-l, -10 times higher than for colchicine; this is reflected in the activation energies for binding which are 14.7 kcal/mol for podophyllotoxin and 20.3 kcal/mol for colchicine. The dissociation rate constant for the tubulin-podophyllotoxin complex is 1.9 h-l, and the affinity constant calculated from the ratio of the rates is close to that obtained by equilibrium measurements. Podophyllotoxin and colchicine are mutually competitive inhibitors.
This can be ascribed to the fact that both compounds have a trimethoxyphenyl ring and analogues of either compound with bulky substituents in their trimethoxyphenyl moiety are unable to inhibit the binding of either of the two ligands.
Tropolone, which inhibits colchicine binding competitively, has no effect on the podophyllotoxinltubulin reaction. Conversely, podophyllotoxin does not influence tropolone binding. Moreover, the tropolone binding site of tubulin does not show the temperature and pH lability of the colchicine and podophyllotoxin domains, hence this lability can be ascribed to the trimethoxyphenyl binding region of tubulin.
Since podophyllotoxin analogues with a modified B ring do not bind, it is concluded that both podophyllotoxin and colchicine each have at least two points of attachment to tubulin and that they share one of them, the binding region of the trimethoxyphenyl moiety. competitively , and this has been ascribed to the fact that both compounds possess a trimethoxyphenyl ring (7). Nevertheless, there are important differences between these ligands.
Thus, the tropolone moiety is an attachment point for colchicine to tubulin (71, and podophyllotoxin is devoid of this ring.
Secondly, colchicine binds slowly, requiring lV2 h to attain equilibrium at 37" (3,(5)(6)(7)(8)(9) and is reversible with difficulty (2, 5, lo), whereas podophyllotoxin is reported to bind and dissociate more rapidly (9,10 In some experiments we used the polymerization method described by Shelanskiet al. (12) or a procedure which combined both of the methods: after one or two cvcles of nolvmerization of rat brain tubulin (12). the nrotein was &polyme;i&d in PMG buffer at 0" for 20 min, applied to a DEAEcellulose column and eluted in the usual fashion (11 -The DEAE-filter paper disc assay for colchicine had to be modified to make it suitable for podophyllotoxin. Two DE81 paper discs (Whatman) were washed with cold PMG buffer (4") by suction, taking care not to dry the paper. Over a period of 1 to 2 min, 100 to 150 ~1 of the sample were applied and were absorbed to filters. The filters were then rinsed four times with 4 ml of cold PMG buffer by mild suction. The discs were counted in 10 ml of Aquasol (New England Nuclear) or Hydromix (Yorktown) to a counting error of ~2%. Identical blanks, lacking only tubulin, amounted to 0.3 to 0.5% of the total radioactivity applied. When 3 or 4 filter papers were used, or when the filters were rinsed with buffer containing lo-' M podophyllotoxin, identical results were obtained. Most of the free podophyllotoxin was removed after the first wash. By contrast, the bound complex remained on the paper and was not washed off with up to seven additional washes as shown in Table I. The formation of podophyllotoxin-tubulin complex, detected by the paper disc method described, is a linear function of protein concentration, as shown in Fig. 1   liver glutamic dehydrogenase, rabbit muscle lactic dehydrogenase, and bovine thyroglobulin.
No chemical modification of the ligand could be detected upon complex formation. Tubulin was incubated for 30 min with ["Hlpodophyllotoxin and the complex was separated by the DEAE-filter paper disc method. The podophyllotoxin was then released from tubulin by heating the DEAE-paper at 60 for 15 min in water. The solution was concentrated and chromatographed on silica gel plates in the four different solvents listed under "Experimental Procedures." In all solvents, more than 97% of the radioactivity moved with authentic podophyllotoxin.

Loss of Binding
Activity of Tub&in-At 37", the colchicine binding activity of uncomplexed tubulin decays in a first order manner with a tliP of 3 to 5 h, and vinblastine and sucrose protect this binding activity (5,8,(21)(22)(23). In Fig. 3, we compare the effect of preincubation of tubulin at 37" on the binding capacity of the protein toward colchicine and podophyllotoxin. The decay of the binding capacity for colcemid was also tested. The t,iz for podophyllotoxin was 5.0 h, for colchicine it was 4.5 h, and for colcemid it was 5.8 h. When the incubation time is added, these tlrL values become, respectively, 5.5 h, 6.0 h, and 6.3 h.
When the preincubation was carried out in the presence of vinblastine or sucrose, the decay of the binding activity for all ligands was much slower (tliz > 14 h in all cases). Vinblastine afforded better protection than sucrose. Heating the protein at temperatures higher than 37" accelerated the decay of the binding activity, and at 60" less than 15 min were needed to abolish completely the ability of tubulin to react with podophyllotoxin or colchicine (Table III).
These results suggested that the lability of colchicine binding resided at least in the trimethoxyphenyl portion of the binding site. They did not, however, shed light on the thermal properties of the tropolone portion of the colchicine site. To investigate this question, we measured the decay in the ability of tubulin to enhance the fluorescence of tropolone. As shown in Table III all at pH 6.7. After the preincubation period the binding activity of tubulin was measured by incubating an aliquot of the sample with the "H-labeled ligands at 37" for 30 min (colcemid and podophyllotoxin), and for 1 h 30 min in the case of colchicine. Colcemid binding was assayed by a method identical to that used for podophyllotoxin. capacity of the protein completely intact whereas colchicine and podophyllotoxin binding were completely abolished. An additional difference between the trimethoxyphenyl and tropolone portions of the colchicine binding site is demonstrated in the pH profile shown in Fig. 2. While podophyllotoxin shows a rather sharp optimum near pH 6.7, the fluorescence of tropolone was unaffected over a pH range of 6.5 to 8.8. It seems clear, therefore, that the high lability of the colchicine binding site is restricted to the trimethoxyphenyl portion.
Binding Parameters at Equilibrium-If the overall binding reaction is where T is tubulin, P is podophyllotoxin, and n is the number of podophyllotoxin binding sites on tubulin, then at equilibrium nKP r=l+KP which is the basis of the Scatchard plot shown in Fig. 4. The number of podophyllotoxin binding sites per tubulin dimer was equal to 0.71 (Fig. 4). The [JH1podophyllotoxin obtained from the complex was 97% pure whereas the mean radiochemical purity of the starting material was 89% (see "Methods").
Thus the impurities did not appear to bind to tubulin, and the corrected value for the stoichiometry is 0.8 mol of podophyllotoxin per mol of tubulin (110,000 daltons). This is identical to the stoichiometry reported for colchicine (3,6,7).
The affinity constant at 37" was 1.6 to 1.8 x 10" Mm' (K,, = l/K = 5.5 to 5.2 x 10m7 M). This should be compared with a K,, value of 7 x 10e7 M obtained for pig brain tubulin (91, and a concentration of 5 x lo-' M required for half-maximal inhibition of axonal transport (24). At lower temperatures, the constant was lower (Fig. 4). A van't Hoff plot of 1ogK uersus the reciprocal of the absolute temperature was linear in the region 22-37" (inset Fig. 4) and yielded an estimate of 4.8 kcal/mol for the standard enthalpy of binding (AH"). The standard free energy (AGO) of binding at 37" was equal to -8.8 kcal/mol and AS" = 43 cal deg-' mall'. The entropy Conditions were adjusted such that <lo% of the reactants were consumed during the intervals when the curves were linear and we have thus assumed that P = P,, and 7' = T,,, where P,, and T,, are the initial concentrations.
When the protein concentration was 190 nM, the rate curves were linear for the first 2 to 3 min of incubation over a concentration of 58 to 188 nM podophyllotoxin (Fig. 5) Table IV; the mean of these values is 3.8 x 10" M-' h-'. These results suggest that podophyllotoxin binding is truly second order under the conditions tested.
Since the association rate constant determined for podophyllotoxin is 10 times higher than that for colchicine (0.36 x 10" M-' h-l (21), 0.41 x 10" M-' h-' (25), or 0.12 to 0.48 x 10" M-' h-l (26)), it was important to measure the activation energies (E,,) for both reactions. A comparison of the temperature effect on colchicine and podophyllotoxin binding, and the Arrhenius plots derived therefrom, is presented in Fig. 6. The activation energy for podophyllotoxin is 14.7 kcal/mol, that for colchicine 20.3 kcal/mol. The higher activation energy for colchicine binding explains, at least in part, its lower association rate.

Dissociation
Rate Constant (k-,)-In view of the discrepancy between the affinity constant (K) determined in equilibrium experiments and the ratio of the association (h,) and dissociation (h-J rate constants for colchicine (21,25,26), it was of interest to determine the Iz,/h-, ratio for podophyllotoxin binding. In order to determine the dissociation of the podophyllotoxin-tubulin complex, the bound podophyllotoxin was separated from the free by passing the sample at 4" through a Sephadex G-75 column (24 x 1.4 cm). The podophyllotoxintubulin complex was diluted -60 times so that reassociation was negligible, and was then incubated at 37" in PMG buffer. As shown in Fig. 7, the t,,4 of dissociation of the complex was -22 min, giving ZL, = 1.9 h-l. The ratio of the rate constant yields an association constant of 2.1 x lo6 Mm', which is in good agreement with the value obtained by equilibrium methods.2 Analogues -Numerous colchicine analogues have been found to block [3H]colchicine binding to tubulin (2,7,8). In most cases it has not been established whether these compete for the A ring or C ring regions or both, or whether binding at one locus necessarily implied occupancy of the whole site. We therefore tested certain analogues for their effect on podophyllotoxin binding on the assumption that only the trimethoxyphenyl moiety of this compound was recognized by the colchicine binding site. If the trimethoxyphenyl ring is a site of interaction of both colchicine and podophyllotoxin with tubulin, it would be expected that analogues with bulky substituents in the trimethoxyphenyl ring are unable to bind to tubulin. This is indeed the case: colchicoside, in which one methoxy group in colchicine ring A is replaced by a sugar, and 4'-demethyl deoxypodophyllotoxin-p-D-glucoside and 4'carbobenzoxy-4'-demethyl podophyllotoxin, which possess bulky substituents in the 4' position of the podophyllotoxin molecule, are totally ineffective in inhibiting colchicine or podophyllotoxin binding (Table V). These analogues also supply evidence that podophyllotoxin binds to tubulin through its trimethoxyphenyl portion. Simple trimethoxyphenyl derivatives do not appear to interact with the colchitine binding site (Table V). However, it has recently been shown that mescaline (3,4,5-trimethoxyphenethylamine) is an antimitotic agent (27). Since podophyllotoxin is known to inhibit colchicine bind-4 For colchicine, the apparent dissociation is the sum of decay of the site plus true dissociation (22). Since t,,* for the podophyllotoxin-tubulin complex j$issociation is 22 min and decay of unoccupied site is about 300 min, the decay factor does not play a significant role in podophyllotoxin dissociation. [3HlPodophyllotoxin and tubulin were incubated at the concentrations stated in the table at 37" and the reaction rate (d[PTlldt) determined in the linear part of the plot of complex formation over time (Fig. 5) As shown in Fig. 8, colchicine was a competitive inhibitor of podophyllotoxin binding. The K, was 2.1 x lo-" M. An important problem for which we have been unable to supply an answer is the discrepancy in ligand affinities for tubulin when these are measured as the Ki or K,) under identical conditions. The Ki values for podophyllotoxin and colchicine are 2.0 x lo-" M and 2.1 x 10e6 M, respectively (7). These differences have been consistently observed. It is conceivable that these discrepancies result from incomplete equilibration.
However, as shown by others as well as by us (3,8,25,28), these reactions were carried out at a time when >90% equilibrium had been attained. Furthermore, since podophyllotoxin binding is more rapid than colchicine binding, failure of equilibration would have opposing effects on Ki and K,l depending on which ligand is labeled. One possible explanation, that the two ligands might interact, seems unlikely since no difference spectra could be elicited between combined and separate solutions of podophyllotoxin and colchicine, and since the mobility of 1 x 10m6 M labeled podophyllotoxin on Sephadex G-10 (in PMG buffer) was not altered by 1 x 10e5 M colchicine. It is possible that the discrepancy between the two constants results from the partial overlap of the binding sites.
Surprisingly, manipulations of the tropolone ring, as shown in isocolchicine, produced analogues that were not able to block podophyllotoxin binding. On the other hand, although both tropolone and methyltropolone bind to the site, as shown by enhancement of fluorescence or blocking of colchicine binding (7), neither has any effect on podophyllotoxin binding. This suggests that the remainder of the podophyllotoxin molecule need not overlap the tropolone domain of the colchicine binding site when the trimethoxyphenyl moiety is binding. This is also shown by the finding that podophyllotoxin did not affect the fluorescence enhancement of tropolone resulting from its binding to tubulin.  analogue. This was necessary in order to maintain a higher molar ratio with these poorly soluble compounds.
The colchicine concentrations were 1.6 x lo-' M for the trimethoxyphenyl analogues and 3 x lo+ M for the podophyllotoxin derivatives with analogue concentrations as above. The plus (+) indicates.that the podophyllotoxin or colchicine binding was <9% than that of a control sample, the minus (-) means that binding was >98% of that of the control.
the same site. However, competitive kinetics are also possible by mutual distortion of the respective binding sites or by steric hindrance from adjacent sites (29 less, certain differences in the binding processes for these inverted with respect to colchicine. Colchicine and podophyllotoxin bind relatively slowly when compared with ANS, which interacts "instantaneously" with tubulin and at a different site (30,31). However, podophyllotoxin is a faster reactant than colchicine, and at 37", its forward rate constant (12,) is 10 times higher than that for colchicine. This appears to be mainly due to the difference in the activation energy which is 20.3 kcal/mol for colchicine and 14.7 kcal/mol for podophyllotoxin.
The structural reason for this difference is, at present, unknown.
Conversely, podophyllotoxin does not affect tropolone binding. The results strongly suggest that the trimethoxyphenyl portion of the colchicine domain is shared by colchicine and podophyllotoxin, whereas the tropolone portion of the site recognizes only colchicine and not podophyllotoxin.
Interestingly enough, the tropolone site does not show the pH and temperature dependence exhibited by podophyllotoxin or colchicine, thus suggesting that these properties are referable to the trimethoxyphenyl binding region. That the trimethoxyphenyl ring is a point of attachment of both colchicine and podophyllotoxin is shown by the observation that analogues with a bulky substituent instead of one of the OCH, groups are unable to inhibit colchicine or podophyllotoxin binding.
Surprisingly, colchicine analogues with a modified C ring and an intact A ring are unable to inhibit podophyllotoxin binding (Table V). These observations suggest that the colchitine molecule is very rigid, such that when the tropolone moiety cannot bind the trimethoxyphenyl portion is sterically hindered from approaching its portion of the binding site. By the same token, if the A ring is hindered then the C ring is prevented from binding as shown by the absence of fluorescence in colchicoside (7). This is in agreement with the crystallographic data for colchicine (32) and colcemid (33).
Like colchicine, podophyllotoxin must also have at least one other binding site to tubulin. This is demonstrated by the marked differences in biological activity of stereoisomers of podophyllotoxin e.g. cis-trczns isomers in the B ring of podophyllotoxin (16). Similarly, succinylation of the l-OH position abolishes the binding activity of podophyllotoxin (9). It remains to be determined whether the tetrahydronaphthol moiety alone will block podophyllotoxin binding and not colchicine binding.
We may conclude, therefore, that colchicine and podophyllotoxin each bind at least at two tubulin domains and have one of these in common, that portion of the site where the trimethoxyphenyl ring interacts.