The Effects of Colchicine Analogues on the Reaction of Tubulin with Iodo [ l 4 C ] acetarnide and N , N ’-Ethylenebis ( iodoacetamide ) *

We have previously found (Ludueña, R. F., and Roach, M. C. (1981b) Biochemistry 20, 4444-4450) that colchicine and podophyllotoxin inhibit the alkylation of tubulin by iodo[14C]acetamide and the formation of an intrachain cross-link in the beta-tubulin subunit by N,N'-ethylenebis(iodoacetamide) (EBI). It was not clear whether these effects were due to conformational changes in tubulin induced by drugs or to direct steric blockage of the sulfhydryl groups involved. In an effort to characterize further these phenomena, we have examined the effects of single-ring and bicyclic analogues of colchicine on the reaction of tubulin with iodo[14C]acetamide and EBI. We have found that neither the A-ring analogues, 3,4,5-trimethoxybenzyl alcohol, 3,4,5-trimethoxybenzaldehyde, 2,3,4-trimethoxybenzaldehyde, and benzaldehyde, nor the C-ring analogues, tropolone and tropolone methyl ether, inhibited alkylation. In contrast, colchicine, podophyllotoxin, and nocodazole and the bicyclic analogues, 5-(2',3',4'-trimethoxyphenyl)-2-methoxytropone and combretastatin, inhibited tubulin alkylation. Since the presence of a bond joining the A and C rings seems to be the determining factor in the suppression of alkylation, it is likely that inhibition by colchicine of the reaction with iodo[14C] acetamide is due largely to a conformational change induced by colchicine. A different pattern was obtained when the effects on cross-link formation by EBI were examined. Here, all the A-ring analogues, the bicyclic analogues, and colchicine, podophyllotoxin, and nocodazole all inhibited formation of the cross-link, whereas the C-ring analogue tropolone methyl ether did not inhibit cross-link formation. Since compounds whose effect on alkylation is markedly different have the same effect on cross-link formation, it is possible that this effect is a steric one and that perhaps the A-ring of colchicine binds to tubulin very close to one of the sulfhydryls involved in the intrachain cross-link formed by EBI in beta-tubulin.

The Effects of Colchicine Analogues on the Reaction of Tubulin with Iodo[ l4C]acetarnide and N,N'-Ethylenebis(iodoacetamide)* (Received for publication, January 13, 1984, andin revised form, November 13, 1984) Mary Carmen Roach, Susan Bane#$, and Richard F. Ludueiiall From the Department of Biochemistry, The University of Texas Health Science Center, San Antonio, Texas 78284 and the 4Department of Biochemistry, Vanderbilt University School of Medicine, Nashville, Tennessee 37232 We have previously found (Ludueiia, R. F., and Roach, M. C. (1981b) Biochemistry 20,[4444][4445][4446][4447][4448][4449][4450] that colchicine and podophyllotoxin inhibit the alkylation of tubulin by iodo["C]acetamide and the formation of an intrachain cross-link in the &tubulin subunit by N,N'-ethylenebis(iodoacetamide) (EBI). It was not clear whether these effects were due to conformational changes in tubulin induced by drugs or to direct steric blockage of the sulfhydryl groups involved. In an effort to characterize further these phenomena, we have examined the effects of single-ring and bicyclic analogues of colchicine on the reaction of tubulin with iodo[14C] acetamide and EBI. We have found that neither the Aring analogues, 3,4,5-trimethoxybenzyl alcohol, 3,4,5-trimethoxybenzaldehyde, 2,3,4-trimethoxybenzaldehyde, and benzaldehyde, nor the C-ring analogues, tropolone and tropolone methyl ether, inhibited alkylation.
In contrast, colchicine, podophyllotoxin, and nocodazole and the bicyclic analogues, 5-(2',3',4'-trimethoxyphenyl)-2-methoxytropone and combretastatin, inhibited tubulin alkylation. Since the presence of a bond joining the A and C rings seems to be the determining factor in the suppression of alkylation, it is likely that inhibition by colchicine of the reaction with iodo[14C] acetamide is due largely to a conformational change induced by colchicine. A different pattern was obtained when the effects on cross-link formation by EBI were examined. Here, all the A-ring analogues, the bicyclic analogues, and colchicine, podophyllotoxin, and nocodazole all inhibited formation of the cross-link, whereas the C-ring analogue tropolone methyl ether did not inhibit cross-link formation. Since compounds whose effect on alkylation is markedly different have the same effect on cross-link formation, it is possible that this effect is a steric one and that perhaps the A-ring of colchicine binds to tubulin very close to one of the sulfhydryls involved in the intrachain cross-link formed by EBI in &tubulin.
Several structurally dissimilar compounds, loosely known as anti-mitotic drugs, have proven very useful in studies on the mechanism of microtubule assembly. One feature common * This work was supported by Grant GM 23476 from the National Institutes of Health and by Grant AQ-726 from the Robert A. Welch Foundation to R. F. L. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. J Present address: Department of Biochemistry, University of Virginia, Charlottesville, VA 22901. ll To whom correspondence should be addressed.
to these compounds is that they bind with relatively high affinity and specificity to the tubulin molecule, the structural subunit of microtubules. Three of these compounds, colchicine, podophyllotoxin, and nocodazole, are competitive inhibitors of each other's binding to tubulin (Hoebeke et al., 1976;Cortese et al., 1977) as well as being potent inhibitors of microtubule assembly (Hoebeke et al., 1976;Olmsted and Borisy, 1973;Wilson et al., 1976). Although binding and inhibition constants have been determined for these three drugs (Borisy and Taylor, 1967;Hoebeke et al., 1976;Cortese et al., 1977) and although there is good evidence that one of them, colchicine, induced conformational changes in the tubulin molecule (Andreu and Timasheff, 198213;Garland, 1978), there is very little definite information about their binding sites on tubulin or about which regions of the tubulin molecule are affected by the drugs. It is not even known to which of the two subunits of tubulin, a or ( 3 , these drugs bind. One approach that has been useful in analyzing the interactions of these drugs with tubulin is the use of analogues of the individual rings of the tricyclic colchicine molecule. Andreu and Timasheff (1982a) have found that analogues of the A and C rings can each inhibit microtubule assembly in uitro, suggesting that colchicine's inhibition of assembly involves the entire binding site. Similarly, Lin and Hamel (1981) have observed that the colchicine-induced stimulation of tubulin's intrinsic GTPase activity requires the A-ring and is not affected by a C-ring analogue.
We previously have examined the interaction of tubulin with drugs by measuring the effects of drugs on the alkylation of tubulin's sulfhydryl groups Roach, 1981a, 1981b). We have reported that colchicine and podophyllotoxin induce a 19-47% reduction in the rate at which these sulfhydryl groups react with i~do['~C]acetamide. We have also reported that EBI,' a bifunctional analogue of iodoacetamide, makes an intrachain cross-link in the @ subunit of tubulin, causing the subunit to migrate, on discontinuous gels containing Na dodecyl sulfate, as a faster-moving band designated @* (Ludueiia and Roach, 1981a). Colchicine and podophyllotoxin inhibit formation of @* by 92-94% (Ludueiia and Roach, 1981b). From these data, however, we could not conclude whether the effects on alkylation were "steric," that is due to binding of a drug directly to a region containing the affected sulfhydryls, or ''allosteric," where the binding of the drug induced a conformational change affecting sulfhydryls located in other regions.
In order to investigate further the interactions of tubulin with these drugs, we have applied the alkylation method to the approach of using analogues of the A and C rings of Alkylation colchicine. We have found that, whereas colchicine, podophyllotoxin, and nocodazole inhibit the reaction of tubulin with i~do['~C]acetamide neither the A-nor the C-ring analogues do so. Since the bicyclic analogue TMPT, which consists of the A and C rings joined by a single bond, suppresses alkylation, as does combretastatin, which consists of the A ring and a C-like ring joined by a 2-carbon bridge, also suppresses alkylation, it is likely that the suppressive effect of colchicine on alkylation is an "allosteric" one where the majority of affected sulfhydryls is not located at the drugbinding sites.
When we examined the effects of these drugs on p* formation by EBI, we found that the A-ring analogues 3,4,5-trimethoxybenzaldehyde, 2,3,4-trimethoxybenzaldehyde, 3,4,5trimethoxybenzyl alcohol, and benzaldehyde all inhibit p* formation, as do colchicine, podophyllotoxin, nocodazole, TPMT, and combretastatin. In contrast, the C-ring analogue, tropolone methyl ether, enhanced p* formation. Since all the compounds which inhibited p* formation contained or consisted of analogues of the A ring of colchicine, it appears that the binding of the A ring to tubulin is the determining factor in inhibiting p* formation. Since colchicine and podophyllotoxin have very different effects on the conformation of tubulin, as measured by their effects on alkylation by i~d o [ '~C ] acetamide, than do the A-ring analogues, it is conceivable that the effect of the A ring on p* formation is due, not to a conformational change, but to direct steric hindrance of the reaction. In other words, it is possible that one of the sulfhydryls involved in p* formation is located at that site on tubulin where the A ring of colchicine binds.

EXPERIMENTAL PROCEDURES
Materials-Tropolone, colchicine, iodoacetamide, rabbit muscle aldolase, Na iodoacetate, fast green FCF, conalbumin, and Coomassie Brilliant Blue were from Sigma. Na iodoacetate was further purified by reprecipitation from acetone. Conalbumin was reduced and carboxymethylated by the method of Crestfield et al. (1963). Acrylamide and N,N'-methylenebis(acry1amide) were from the Eastman Kodak Co., Rochester, NY. Acrylamide solutions were routinely filtered through charcoal. Na dodecyl sulfate was from the Accurate Chemical and Scientific Corporation, Arlington Heights, IL. EBI was synthesized as previously described (Ludueiia and Roach, 1981a). Vinblastine sulfate was a kind gift from Lilly. Podophyllotoxin, nocodazole, 2,3,4-trimethoxybenzaldehyde, 3,4,5-trimethoxybenzaldehyde, 3,4,5trimethoxybenzyl alcohol, benzaldehyde, and acetaldehyde were from Aldrich. I~do[l-'~C]acetamide was from Amersham Corp. It was diluted with unlabeled iodoacetamide and its specific activity determined as previously described (Ludueiia and Roach, 1981a). [ ring A-4-3H]Colchicine was also from Amersham Corp. Tropolone methyl ether was synthesized by the method of Nozoe et al. (1951) and recrystallized from ether. Its structure was confirmed and its purity assayed by melting point determination, nuclear magnetic resonance, and thin layer chromatography on silica gel plates in methylene ch1oride:methanol (9:l). No contamination was detectable by any technique. Batches of T P M T were the kind gifts of Dr. Thomas J. Fitzgerald, Florida Agricultural and Mechanical University, Tallahassee, FL, and of Dr. James Lee, St. Louis University, St. Louis, MO. Combretastatin was the kind gift of Dr. George Pettit, Arizona State University, Tempe, AZ. Microtubule protein was purified from bovine cerebra, and tubulin was purified from microtubule protein by phosphocellulose chromatography according to the method of Fellous et al. (1977) as previously described (Ludueiia and Roach, 1981a). Mercaptoethanol was omitted from the buffer of Fellous et al. (1977) when tubulin was purified for use in alkylation experiments. All experiments were performed in this buffer. ' Reaction with Zodof4C]acetarnide-Tubulin samples were alkylated with iodo["C]acetamide and then precipitated with trichloroa- The buffer consisted of 0.1 mM 2-(N-morpholino)ethanesulfonic acid, pH 6.4, 1 mM ethylene glycol bis(P-aminoethyl ether)-N,N,N', N'-tetraacetic acid, 0.1 mM EDTA, 0 . 5 m~ MgC12, and 1 mM GTP. cetic acid and the extent of alkylation of the tubulin determined by the method of Ludueiia and Roach (1981~). The binding of [3H] colchicine to tubulin was determined by the filter disc method of Borisy (1972). Protein concentrations were determined by the method of Lowry et al. (1951). Colchicine was dissolved in MES buffer (Fellous et al., 1977) and podophyllotoxin, nocodazole, TPMT, and the analogues of the A and C rings of colchicine were dissolved in dimethyl sulfoxide immediately prior to use; in experiments where these compounds were used, control samples always contained equivalent concentrations of dimethyl sulfoxide.
Reaction with EBI-Tubulin samples were reacted with EBI and then reduced and carboxymethylated (Crestfield et al., 1963) and run on 6% polyacrylamide gels (Laemmli, 1970) which were stained and scanned as described previously (Ludueiia and Roach, 1981a). The areas of the peaks were determined by planimetry. In most experiments where EBI was used, the tubulin samples contained reduced and carboxymethylated conalbumin to permit estimation of the yields of cross-linked aggregated tubulin and of @*. Based on the results of Ludueiia et al. (1982), p* was assumed to derive only from p1 and not from 0 2 and was determined by the following equation, (1) where @*" and C, are the areas on the gel scan of sample n of the p' and conalbumin peaks, respectively, and RBI is the ratio of the p1 subunit to conalbumin in a sample that had not been reacted with EBI. The percentage, designated % @,res, of p1 that was residual, that is that had not reacted with EBI to generate a cross-linked product of altered mobility, was calculated according to the formula, where Pln is the area of on the gel scan of sample n. The percentage, designated % Plagg, of the population of the p1 subunit in a sample n that had reacted with EBI to form a cross-linked aggregate was calculated from the following equation.
Similarly, the percentages of the populations of the p2 or a subunits that had formed a cross-linked aggregate (% bzagg or aagg, respectively) were determined from the equations, (4) and % aagg = [Ian/(RaCn)] X 100% (5) where p2. and a, are the areas of the p 2 and a peaks, respectively, on the gel scan of sample n, and RBz and R, are the ratios of the areas of p2 and a, respectively, to conalbumin on a gel of a control sample not treated with EBI.
In preliminary experiments it was observed that tubulin could spontaneously aggregate to a small extent upon prolonged incubation even in the absence of EBI. In order to minimize this effect, the control samples that had no EBI routinely contained 20 WM vinblastine. In a few EBI-treated samples, where the ratio of + @* to C was unusually high, the % Plagg was a negative number. Whether this was due to normal statistical fluctuation or to distortions in the gel was not always clear. These samples were considered unreliable and were not used in any calculation of %@'res or %@*.

Effect of Colchicine Analogues on the Alkylation of Tubulin by l~do['~C]acetarnide-Some of the experiments involving
the A-ring analogues of colchicine were done using 3,4,5trimethoxybenzaldehyde, a compound previously shown by Lin and Hamel (19810 to mimic colchicine's enhancement of tubulin's intrinsic GTPase activity. The analogues 2,3,4-trimethoxybenzaldehyde, 3,4,5-trimethoxybenzyl alcohol, and benzaldehyde were used as well. As expected for colchicine analogues, 3,4,5-trimethoxybenzaldehyde, 3,4,5-trimethoxybenzyl alcohol, and benzaldehyde inhibited the binding of [3H]colchicine to tubulin (Table I). All four compounds were capable of enhancing the reaction of tubulin with i~d o [ '~C ] acetamide (Table 11), although the effect of 3,4,5-trimethoxybenzyl alcohol was significantly smaller than that of the   at 37 "C. All incubations were done in quadruplicate. Colchicine binding to tubulin was then assayed by the filter disc method of Borisy (1972 other two A-ring analogues. That this effect was specific for tubulin was suggested by the fact that 30 mM 3,4,5-trimethoxybenzaldehyde had no effect on the alkylation of aldolase by i~do['~C]acetamide. In an experiment where triplicate samples of aldolase (1.13 mg/ml) were incubated with 1.36 mM i~do['~C]acetamide for 60 min at 37 "C, the label incorporated into the aldolase in the presence and absence of 3,4,5-trimethoxybenzaldehyde was 0.64 f 0.04 and 0.58 f 0.03 mol/ mol, respectively.
It was conceivable that larger increases in alkylation caused by the aldehyde-containing analogues were due in some way to formation of a Schiff base with tubulin and a subsequent conformational change. This would be an effect that need not involve the colchicine-binding site. In order to test this, the effect of acetaldehyde was examined, since this compound would have the aldehyde moiety without any resemblance to colchicine. As can be seen from Table I, a 1-h incubation of tubulin with acetaldehyde had no effect on the binding of [3H] colchicine, whereas all of the A-ring analogues inhibited colchicine binding under the same conditions.
After 3 h of incubation, however, acetaldehyde inhibited colchicine binding by 35%, much less than the effect of the A-ring analogues. It is possible that the long-term effect of acetaldehyde was due to Schiff base formation requiring 3 h to show an effect. Nevertheless, it is striking that the A-ring analogues inhibit colchicine binding after only an hour of incubation, indicating that they are binding to the colchicine site and suggesting that the enhancing effect on alkylation shown by 3,4,5-tri- " Aliquots (250 p1) of tubulin (0.66 mg/ml) were incubated in triplicate for 60 min at 37 "C in the presence of the indicated compounds. They were then reacted with 1.34 mM iodo["C]acetamide (0.49 Ci/mol) for 60 min at 37 "C, and the incorporation of I 4 C label into each sample was calculated as previously described (Ludueiia and Roach, 1981~).
Standard deviation. Experiments 2 and 3 were performed under the same conditions as Experiment 1 except the specific activity of the i~do['~C]acetamide was 0.31 Ci/mol.
Number of aliquots used. e Experiment 4 was performed under the same conditions as Experiment 1 except the specific activity of the i~do['~]acetamide was 0.51 Ci/mol. methoxybenzaldehyde and benzaldehyde must be in some way a consequence of their binding to the colchicine site.
The enhancement of alkylation of tubulin by 3,4,5-trimethoxybenzaldehyde was apparently due to a simple increase in the rate of alkylation of tubulin (Fig. 1). The effect required a 3,4,5-trimethoxybenzaldehyde concentration of at least 2 mM and was observed in the presence and absence of vinblastine (Fig. 2). As previously described (Ludueiia and Roach, 1981b), vinblastine is a potent inhibitor of the alkylation of tubulin by i~do['~C]acetamide. It binds to tubulin a t a site or sites distinct from that where colchicine binds (Bryan, 1972). 3,4,5-Trimethoxybenzaldehyde, at a concentration of 4 mM, begins to reverse vinblastine's suppression of alkylation; only at 30 mM, however, does the enhancement of alkylation by 3,4,5-trimethoxybenzaldehyde exceed that observed with tubulin in the absence of either vinblastine or 3,4,5-trimethoxybenzaldehyde (Fig. 2).
Tropolone significantly enhanced the alkylation of tubulin by i~do['~C]acetamide, although concentrations of 10 mM or higher were required (Table 11). At these concentrations, tropolone did not appear to have a significant effect on the alkylation of aldolase. In an experiment where duplicate samples of aldolase were alkylated with 1.36 mM iodo['"C]aceta- . 1. Effect of 3,4,54rimethoxybenzaldehyde on the alkylation of tubulin by iodo["C]acetamide. Aliquots (250 pl) of tubulin (0.66 mg/ml) were incubated at 37 "C for the indicated times with 1.36 mM i~do['~C]acetamide (0.31 Ci/mol) in the presence (0) or absence (0) of 30 mM 3,4,5-trimethoxybenzaldehyde. The radioactivity incorporated into tubulin was measured as described by Ludueiia and Roach (1981~). Each time point represents a separate incubation of triplicate samples. Standard deviations are shown except where they are smaller than the symbol. Tropolone methyl ether caused a more moderate enhancement of alkylation although the effects were only visible at concentrations of 10 mM or higher (Fig. 2). At this concentration range, tropolone methyl ether inhibited colchicine binding ( Table I). The effect of tropolone methyl ether was apparently due to an increase in the rate of alkylation of tubulin by iod~[*~C]acetamide (not shown). When tropolone methyl ether and 3,4,5-trimethoxybenzaldehyde were tested in combination, their effects on incorporation of iodo[ 14C]acetamide were indistinguishable from those of 3,4,5-trimethoxybenzaldehyde by itself (Fig. 2).
In order to see what effect linking the A and C rings would have on alkylation, experiments were done with TPMT and combretastatin. TPMT was able to inhibit the alkylation of tubulin by i~do['~C]acetamide (Fig. 2) at concentrations where it also inhibited binding of [3H]colchicine to tubulin ( Table I). Half-maximal inhibition was obtained at about 2-10 PM TMPT. The inhibition of alkylation was due to a decrease in the rate of reaction of tubulin with i~do['~C] acetamide (not shown). As can be seen in Table 11, combretastatin also inhibits alkylation. This bicyclic compound is an analogue of colchicine lacking the acetamide group on the B ring and the covalent bond joining the A and C rings; combretastatin also has colchicine's C ring replaced by a phenylene group with methoxy and hydroxyl groups at positions equivalent, respectively, to those where a methoxy and ketone group occur on the C ring of colchicine (Pettit et al., 1982;Hamel and Lin, 1983). Combretastatin thus consists of the A ring and a C-like ring joined by a 2-carbon bridge and, in terms of the B ring, is complementary to TPMT. This inhibition of alkylation is in marked contrast to the effects of the individual A-and C-ring analogues, which enhance alkylation. Even when 3,4,5-trimethoxybenzaldehyde and tropolone methyl ether were tested together, they failed to inhibit alkylation (Fig. Z), suggesting that a bond or bridge, joining the A and C rings, is critical in determining the effect of the compound on the alkylation of tubulin by i~do['~C]acetamide.

: -
As one might expect, analogues which strongly enhanced alkylation also enhanced formation of this aggregate, which made the decreased yield of @* difficult to interpret. For example, in the presence of EBI and 30 mM 3,4,5-trimethoxybenzaldehyde about 88% of a, 82% of Dl, and 74% of Pz were incorporated into this aggregate. It was, therefore, not clear if the analogue inhibited @* formation by directly blocking that particular reaction or by enhancing a competing reaction. Fig. 3, however, suggests that the former may be the case, since 3,4,5-trimethoxybenzaldehyde slowed down the rate of disappearance of @*.
In order to minimize the contribution of nonspecific crosslinking, a similar experiment was done in the presence of vinblastine, a drug known to inhibit nonspecific cross-linking by EBI, while enhancing formation of @* (Ludueiia and Roach, 1981b). When the effect of 3,4,5-trimethoxybenzaldehyde on examined (Figs. 3-5), it was found that the yield of @* was suppressed by 80% at a concentration of 15 mM 3,4,5-trimethoxybenzaldehyde. Even at concentrations below 15 mM, 3,4,5-trimethoxybenzaldehyde significantly inhibited @* formation, half-maximal inhibition being obtained with 2 mM 3,4,5-trimethoxybenzaldehyde. Most important, as the concentration of 3,4,5-trimethoxybenzaldehyde increased from 0 to 15 mM, the yield of residual (noncross-linked) Dl increased from 22 to 47% and that of @* decreased from 56 to 11% (Fig.   5), indicating that 3,4,5-trimethoxybenzaldehyde was actually inhibiting formation of the @* cross-link, rather than merely increasing the transformation of p1 into a competing structure such as the cross-linked aggregate.
As shown in Fig. .3, in the presence of vinblastine, 3,4,5trimethoxybenzaldehyde inhibited the rate of formation of @* and decreased the rate of disappearance of pl, consistent with what was observed in the absence of vinblastine. The effects of 3,4,5-trimethoxybenzyl alcohol and benzaldehyde on the reaction of EBI with tubulin were also examined (Table 111). As can be seen, these analogues significantly inhibited @* formation by EBI while concomitantly increasing the yield of &res. Since 3,4,5-trimethoxybenzyl alcohol and benzaldehyde had little effect on nonspecific cross-linking, as represented by Blagg, it is likely that the decreased yield of @* was due directly to inhibition of cross-link formation.
Interestingly, acetaldehyde inhibited @* formation and en-  'The experimental conditions were identical to those used in Experiment 1, except that all samples contained 20 p~ vinblastine and the incubation was at 28 "C.
Experimental conditions were identical to those used in Experiment 1. excent that all incubations were in auadrudicate and all @* formation by EBI in the presence of vinblastine was samples contained 20 p~ vinblastine. Analogues and Tubulin Alkylation hanced the yield of Plres, suggesting that it was acting as a genuine inhibitor of the formation of the cross-link.
As shown in Table 11, tropolone significantly enhanced alkylation of tubulin by i~do['~C]acetamide. Tropolone also enhanced nonspecific cross-linking by EBI to a very large extent. For example, in the presence of 40 mM tropolone, 97% of a , 92% of pz, and 83% of p1 disappeared from the gel. Even the addition of vinblastine did not significantly diminish this effect of tropolone. In one experiment, for instance, in the presence of 20 pM vinblastine, 20 mM tropolone caused the disappearance from the gel of 63% of a , 54% of p2, and 65% of pl. In contrast to the A-ring analogues, however, tropolone did not induce as great a decrease in the yield of @* relative to as did the A-ring analogues. For example, 20 mM tropolone generated @*/PI ratios of 1.15 and 0.93 in the presence and absence, respectively, of 20 mM vinblastine. Nevertheless, the strong enhancement of nonspecific cross-linking by tropolone made this compound an unsatisfactory choice for examining the effect of C-ring analogues on p* formation.
Tropolone methyl ether caused only a small increase in nonspecific cross-linking (Fig. 6). Interestingly, tropolone methyl ether did not significantly diminish the yield of @*, even at concentrations where it enhanced alkylation by i~do['~C]acetamide. In fact, tropolone methyl ether actually increased the yield of @* relative to p1 (Figs. 3, 4, and 6). For example, when the tropolone methyl ether concentration increased from 0 to 20 mM, the yield of residual PI decreased by 17% while the yields of p* and aggregated PI increased by 9 and 8%, respectively (Fig. 6). In a similar experiment done in the presence of 20 p~ vinblastine, an increase of the tropolone methyl ether concentration from 0 to 20 mM caused a 9% decrease in the yield of residual Dl, while the yields of @* and aggregated increased by 7 and 3%, respectively. These results suggest that the increased yield of @* relative to  ' (0, A, A), the % &agg (0, V), and the % plres caused by tropolone methyl ether is due at least in part to the direct enhancement of formation of the cross-link which generates p* as well as by enhancement of nonspecific crosslinking. Only at high concentrations (40 mM) did tropolone methyl ether induce a decrease in the yield of p*, this being accompanied by a large increase in the yield of cross-linked aggregated (Fig. 6). When tropolone methyl ether and 3,4,5trimethoxybenzaldehyde were tested together, the effect on @* formation was the same as that of 3,4,5-trimethoxybenzaldehyde alone (Fig. 6). The enhancing effect of tropolone methyl ether on @* formation appeared to have a complex time dependence, being less pronounced at longer reaction times. However, tropolone methyl ether strongly enhanced the rate of disappearance of Dl, suggesting that at longer reaction times (after 30 min), tropolone methyl ether enhanced the formation of aggregated cross-linked p1 at the expense of @* (Fig. 2).
When tested in the presence of EBI, T P M T markedly decreased @* formation (Fig. 7). The effect was half-maximal at a T P M T concentration of about 8 p~. Like colchicine (Ludueiia and Roach, 1981b), T P M T also induced a decrease in nonspecific cross-linking. The suppression of @* formation by T P M T appeared to be a simple inhibition of the reaction rate (Fig. 3) and was accompanied by a strong inhibition of the rate of disappearance of pl . In similar fashion, combretastatin inhibition both the formation of @* and the disappearance of p1 (Table 111).

Effect of Colchicine Analogues on the Alkylation of Tubulin by Zodo['*C]acetanide-It
is clear from the results presented in Table I1 and Figs. 1 and 2 that the analogues of the individual A and C rings of colchicine either enhance or at least do not inhibit the alkylation of tubulin by iodo[14C] acetamide at concentrations where they are binding specifically to tubulin as shown by their inhibition of colchicine binding (Table I). This result is in marked contrast to the effects of colchicine, podophyllotoxin, nocodazole, TPMT, and combretastatin, all of which inhibit the reaction of tubulin with iodo["C]acetamide. It may be argued that 3,4,5-trimethoxybenzaldehyde, 2,3,4-trimethoxybenzaldehyde, and benzaldehyde could react covalently with tubulin at some site distinct from the colchicine-binding site to form a Schiff base and that such a reaction could affect the conformation of the tubulin molecule by a unique mechanism that would not involve the colchicine-binding site. By this argument the strong enhancement of alkylation caused by these A-ring analogues could conceivably be a result of such a conformational change. However, acetaldehyde, which has an aldehyde moiety but no resemblance to colchicine, has only a small effect on alkylation (Table 11). On the other hand, the A-ring analogue, 3,4,5-trimethoxybenzyl alcohol, which lacks the aldehyde moiety, also has little effect on alkylation. It is conceivable that the combination of the resemblance to colchicine and the presence of the aldehyde moiety may be responsible for the strong enhancement of alkylation, perhaps as a result of binding at the colchicine site followed by formation of a Schiff base with a neighboring favorably oriented amino group. It is interesting that much stronger inhibition of colchicine binding by the A-ring analogues was observed with 3,4,5-trimethoxybenzaldehyde and benzaldehyde than with 3,4,5-trimethoxybenzyl alcohol or acetaldehyde. In fact, acetaldehyde only became inhibitory after a long incubation ( Table I). The important point, however, is not to what extent the A-and C-ring analogues enhance alkylation of tubulin by i~do['~C]acetamide but that they do not inhibit it, in contrast to colchicine and the bifunctional analogues which do inhibit it.
It is interesting that TPMT suppresses alkylation because structurally T M P T consists of the A and C rings joined by a covalent bond. The fact that the combination of 3,4,5-trimethoxybenzaldehyde and tropolone methyl ether also enhances alkylation (Fig. 2) suggests that the fact of the A and C rings being joined, as they are in TPMT, combretastatin, and colchicine, is all important in determining whether alkylation will be inhibited or enhanced. If the inhibition of alkylation by colchicine and TPMT is purely a steric effect, that is due to direct blockage of the reacting sulfhydryls, then the sulfhydryls in question would have to be located precisely in that portion of the binding site which is closest to the bond joining the A and C rings. Although the precise number of sulfhydryls affected by colchicine has not been determined, Ludueiia and Roach (198lb) have shown that these sulfhydryls are located on both the a and /3 subunits. In other words, for the steric model to be true, the bond joining the A and C rings would not only have to be located close to the affected sulfhydryl, it would also have to straddle the a l p interface.
The fact that combretastatin inhibits alkylation makes this model less likely, however, because the bond which joins the A and C rings in T P M T is absent in combretastatin. Instead the A ring and the ring equivalent to the C ring are joined by a 2-carbon bridge similar to the outer part of colchicine's B ring. It is difficult to explain by a simple steric model how single A-and C-ring analogues, even in combination, fail to suppress alkylation while any compound in which the two rings are connected, no matter how, will suppress alkylation.
The results presented in this paper lend themselves to another model, an allosteric one, where colchicine binding to tubulin induces a conformational change in the molecule which affects several sulfhydryl groups on both subunits. The conformational change would be dependent upon the presence of the bond joining the A and C rings. By this model, the single ring analogues of colchicine would either induce conformational changes in tubulin or destabilize its conformation in such a way as to enhance the alkylation of tubulin. Such a model is in apparent contradiction to the results of Andreu and Timasheff (1982b), who find that tropolone methyl ether and colchicine affect the circular dichroism spectrum of tubulin in similar fashions, while podophyllotoxin, considered here as an A-ring analogue, has a different effect. It must be considered, however, that a slight shift in the relative positions of two domains in tubulin induced by the binding of a ligand could conceivably have little or no effect on the circular dichroism spectrum of the molecule but could greatly alter the accessibility of certain sulfhydryl groups to alkylating agents. Along the same lines, Andreu and Timasheff (1982a) find that the observed free energies of tubulin binding to colchicine, tropolone methyl ether, and the A-ring analogue N-acetylmescaline are consistent with a model whereby the bindings of the A-and C-ring moieties of colchicine to tubulin are not altered by the existence of the covalent bond joining the two rings. Our results, in contrast, would imply that the existence of this bond causes changes in the accessibility of certain sulfhydryl groups which are very different from those caused by either of the single ring analogues. Again it is conceivable that a very small change in the overall conformation of the protein could greatly alter the accessibility of certain sulfhydryl groups but that the free energy of such a change could be small enough to be within experimental error of the free energies estimated by Andreu and Timasheff (1982a). It is probable, therefore, that the conformational change which we postulate to account for colchicine's inhibition of alkylation is just a small part of the larger conformational change which Andreu and Timasheff (1982a) propose in which the binding of the C-ring moiety of colchicine facilitates the binding of the A-ring moiety.
It is interesting that Andreu et al. (1984) have recently shown that TPMT affects tubulin's intrinsic fluorescence, circular dichroism, and GTPase activity in a manner similar to that of colchicine. They propose that the conformation of the tubulin molecule when it is bound to TPMT is similar to its conformation when it is bound to colchicine. This is consistent with our observation that TPMT and colchicine both inhibit the alkylation of tubulin by i~do['~C]acetamide.
As shown in Table 11, podophyllotoxin and nocodazole also inhibit alkylation by i~do['~C]acetamide. It is possible that these effects are also due to putative conformational changes induced by these drugs, but the possibility cannot be eliminated that there is a steric inhibition of alkylation due to those portions of podophyllotoxin and nocodazole which bind to areas on the tubulin molecule where colchicine does not bind. In order to investigate this possibility it would be necessary to examine the effects of analogues of individual portions of these drugs, such as the tetracyclic component of podophyllotoxin.
The Effect of Colchicine Analogues on the Formation of the /3* Cross-link by EBI-Previously published work (Ludueiia and Roach, 1981b) and the data shown here indicate that colchicine, podophyllotoxin, nocodazole, TPMT, and combretastatin are potent inhibitors of /3* formation by EBI, suggesting that these compounds, when they bind to tubulin, have a strong effect on at least one of the sulfhydryl groups involved in /3* formation. When the effects of the analogues of the individual colchicine rings are examined, it is seen that only the A-ring analogues can suppress /3* formation. Of the two C-ring analogues used in this study, tropolone induced too much nonspecific cross-linking to allow accurate estimation of a direct effect on p* formation, while the other, tropolone methyl ether, directly enhanced formation of p*. In contrast, the A-ring analogues 3,4,5-trimethoxybenzyaldehyde, 2,3,4-trimethoxybenzaldehyde, 3,4,5-trimethoxybenzyl Colchicine Analogues and Tubulin Alkylation alcohol, and benzaldehyde suppressed /?* formation. In the absence of vinblastine, the A-ring analogues caused a large increase in the formation of cross-linked aggregate by EBI, an effect which did not allow demonstration of whether /?* formation was decreased by true inhibition of cross-linking or by formation of a competing cross-linked product. However, in the presence of vinblastine, which greatly diminishes nonspecific alkylation (Ludueiia and Roach, 1981b) the Aring analogues induced a lowered yield of /?* and a higher yield of residual (noncross-linked) PI. Apparently, therefore, these compounds also act by specifically inhibiting the reaction with EBI of at least one of the sulfhydryl groups involved in p* formation. It thus seems that the ability to influence these particular sulfhydryl groups is restricted to those compounds which contain a trimethoxyphenyl ring or, in the case of nocodazole, an equivalently placed phenylene moiety. As shown above, colchicine, podophyllotoxin, nocodazole, TPMT, and combretastatin are potent inhibitors of the alkylation of tubulin by i~do['~C]acetamide. The evidence suggests, particularly in the case of colchicine, that this effect is conformational in nature, that is the affected sulfhydryls are not likely to be located at the drug-binding site. In contrast, the A-ring analogues enhance alkylation of sulfhydryl groups which again must not be located at their binding site. In addition, Andreu and Timasheff (1982b) have shown that podophyllotoxin and colchicine induce different conformational effects on the tubulin molecule. The fact that all of these compounds, whose conformational effects on the tubulin molecule appear to vary considerably, all affect the same sulfhydryl or sulfhydryls the same way, suggests that this latter effect is not due to a conformational change induced by the ligand but to direct hindrance of the sulfhydryl or sulfhydryls by the ligand. In other words, it is likely that these sulfhydryls are located at or near to the region on the tubulin molecule where the A ring of colchicine binds, this presumably being the place where the binding sites of podophyllotoxin and nocodazole overlap colchicine's (Cortese et al., 1977;Lin and Hamel, 1981). It is conceivable, however, that even though the overall conformational effects on tubulin of different ligands may be very different, there may be a specific region of tubulin where these effects are the same, and it is certainly possible that one or both of the sulfhydryl groups involved in p* formation may be located in this region. Another complicating factor is that whereas ( 3 ' formation requires two sulfhydryls, inhibition of p* formation requires that only one sulfhydryl be prevented from reacting with EBI. It is conceivable, though unlikely, that the different ligands act by suppressing the reactivity of different sulfhydryls. If this were the case, then our argument based on a common mechanism of action of different ligands would be less convincing and the likelihood of the effects being allosteric rather than steric would be higher. It is interesting that acetaldehyde should also suppress /?* formation (Table 111), since this compound does not bind at the colchicine-binding site (Table I). This result raises the possibility that inhibition of /?* formation is due to a reaction of an aldehyde group with one of the reactive sulfhydryls, which need not involve the colchicine-binding site. However, the fact that 3,4,5-trimethoxybenzyl alcohol is a good inhibitor of p* formation (Table 111) suggests that the presence of the aldehyde group is not necessary for this to happen. Conceivably, acetaldehyde forms a Schiff base elsewhere in the tubulin molecule which coincidentally prevents p* formation in a different way, perhaps by being located between the two /?* sulfhydryls. The tentative model advanced here, that the sulfhydryls involved in p* formation are located at the site where the A ring of colchicine binds, has three interesting corollaries. The first and most obvious is that these sulfhydryls are on the p subunit of tubulin and that the colchicine-binding site is located, at least in part, on (3. This is an apparent contradiction to the preliminary results of Barnes et al. (1982) who find that a photoaffinity analogue of colchicine can be covalently linked to the a chain. However, this analogue has its photoreactive portion attached to the B ring and the fact that it cross-links to a may simply mean that the B ring is no further away from the a chain than the length of the photoreactive moiety. Such a conclusion is in no way contradictory to a model where the A ring binds to /?. It is also possible that the colchicine-binding site overlaps both the a and /? subunits. A second corollary of the model derives from the observations by Palanivelu and Ludueiia (1982) and Ludueiia et al. (1982) that one or both of the sulfhydryls involved in p* formation is critical for microtubule assembly and that a molecule of tubulin that has reacted with EBI to form /?* cannot assemble. If the model advanced here is correct and one or both of these sulfhydryls are located at the site where colchicine, podophyllotoxin, and nocodazole bind, then we may deduce that this region of the tubulin molecule must have a critical role in determining whether a molecule of tubulin may polymerize. Polymerization to microtubules would be inhibited either by a reagent which reacts with the sulfhydryls at this site or by a drug, such as colchicine, which covers them. Whether this site serves as a tubulin-tubulin binding site in microtubule assembly or whether the events at this site affect a more distant tubulin-tubulin interaction site is unclear.
A third corollary of this model derives from the observation in Figs. 6 and 8 that tropolone methyl ether enhances /?* formation. If the sulfhydryl groups involved in @* formation are located where the A ring binds, then it is probable that the binding of tropolone methyl ether to tubulin causes a conformational change which affects the A-ring-binding site, perhaps moving it so that the orientation of the sulfhydryl groups is altered in such a way as to facilitate p* formation. This is very similar to what Andreu and Timasheff (1982a) have proposed, namely that the C-ring moiety of colchicine binds first and that it induces a conformational change which moves the A-ring-binding site so as to facilitate the binding of the A-ring moiety.
In summary, it appears that the sulfhydryl groups of tubulin may be sensitive indicators of the conformational changes induced in the tubulin molecule by the binding of ligands, giving information at a resolution as high if not higher than more standard physical biochemical techniques. It also seems that there is a region on the tubulin molecule containing the p* sulfhydryls and that this region is strongly affected by events at the site where the A ring of colchicine binds. It is likely, but not definite, that these two sites are one and the same. Experiments, currently in progress, to locate the p* sulfhydryls in the primary sequence of tubulin and to identify the sulfhydryl groups which have their alkylation affected by the binding of colchicine should greatly increase our understanding of the mechanisms of action of colchicine and other anti-mitotic drugs.