A Photoaffinity Derivative of Colchicine: 6’-(4’-Azido-2’-nitrophenylamino)hexanoyldeacetylcolchicine PHOTOLABELING AND LOCATION OF THE COLCHICINE-BINDING SITE ON THE a-SUBUNIT OF TUBULIN*

A photoaffinity analog of colchicine, 6-(4’-azido-2‘-nitrophenylamino)hexanoyldeacetylcolchicine, was synthesized by reacting deacetylcolchicine or [3H]de-acetylcolchicine with N-succinimidyl-6-(4’-azido-2’-nitropheny1amino)hexanoate. Homogeneity of the pho- toaffinity analog was established by thin-layer chromatography and high-pressure liquid chromatogra- phy. The structure of the photoaffinity analog was determined by ‘H and 13C NMR, infrared and ultravi-olet-visible spectroscopies, and elemental analysis. Binding of 6-(4’-azido-2‘-nitrophenylamino)hexano- yldeacetylcolchicine to bovine renal tubulin was measured by competition with [3H]colchicine. The value of the apparent Ki for the photoaffinity analog

Essential to understanding the reaction of colchicine with tubulin is knowledge of the properties and location of the colchicine-binding site on tubulin. Such knowledge should also facilitate elucidation of interactions among colchicine and other ligands which bind to tubulin, the role of colchicine in disassembly of microtubules, and isolation and identification of a possible endogenous ligand. Photoaffinity labeling has been used extensively in the selective labeling of receptor sites in a large variety of biological systems, especially with aryl azides as the photolabile moiety (1). Photogenerated GM30665 to R. F. W. Portions of this work are for partial fulfillment of the requirements for a masters degree by M. J. A. and for a doctoral degree by L. J. F. The costs of publication of this article were defrayed be hereby marked "advertisement" in accordance with 18 U.S.C.
in part by the payment of page charges. This article must therefore Section 1734 solely to indicate this fact. reagents are potentially highly reactive and can be activated in situ under mild conditions. These are advantageous properties for labeling binding sites located in hydrophobic regions of proteins (2).
In this report we describe the synthesis, chemical characterization, and binding properties of 6-(4'-azido-2'-nitrophen-y1amino)hexanoyldeacetylcolchicine (ANPAH-CLC'), a new photoaffinity derivative of colchicine. We report also the photolabeling and subunit localization of the colchicine-binding site of renal tubulin with this analog.
Previous studies indicate that the interaction between colchicine and tubulin is primarily nonionic, noncovalent, and occurs in a hydrophobic domain of tubulin (3). Two different approaches have been taken to specifically locate the colchicine-binding site on tubulin with reactive analogs of colchicine. Bryan and co-workers synthesized two photoaffinity analogs of colchicine, but photolabeling of tubulin was unsuccessful with both analogs (4, 5). Chlorocyanoethyl colchicine and diazomalonyl colchicine were photolyzed in the presence of brain tubulin, but neither a-nor 8-tubulin were labeled. Rather, a protein with a molecular weight of 16,500 was labeled, and the relationship, if any, of this protein to tubulin was unknown. Schmitt and Atlas (6) used bromocolchicine, a compound with nonspecific alkylating reactivity, to label tubulin. They concluded the colchicine-binding site was on the or-subunit, although both or-and &subunits were labeled. Ludueiia has reviewed the problems associated with such general alkylating compounds (7).

RESULTS
Synthesis of 6-(4'-Azido-2'-nitrophenylamino)hexanoyldeacetylcolchicine-The photoaffinity derivative of CLC was synthesized by acylation of the amino group of DAC with SANPAH. DAC (0.3 mol) was reacted with SANPAH (0.3 mol) in 20 ml of acetone under Nz as shown in Equation 1. The solution was stirred for 48 h at room temperature. The reaction was monitored with time by TLC and the reaction was allowed to proceed until DAC was undetectable. Solvent was evaporated, and the resultant solid was dissolved in a minimum volume of methylene chloride. The mixture was purified by preparative TLC on Whatman PLK5F linear K silica gel plates (20 X 20 cm; 1000 p ) in benzene-methanol (3:l). The product was isolated by elution with methanol and recrystallized from methylene chloride-diethyl ether. The yield of ANPAH-CLC from DAC and SANPAH was 71%.
Characterization of ANPAH-CLC-ANPAH-CLC was homogeneous based on TLC on silica gel in benzene:methanol (3:l) (values of RF: ANPAH-CLC, 0.5; DAC, 0.33; SANPAH, 0.95) and in methanol (values of RF: ANPAH-CLC, 0.68; DAC, 0.40; SANPAH, 0.70). ANPAH-CLC exhibited a single peak when subjected to reversed-phase HPLC. A contamination of 0.5% (w/w) of DAC or SANPAH in the purified ANPAH-CLC preparation would have been detectable by HPLC (Fig. 1A). The materials which eluted adjacent to DAC in the HPLC (Fig. 1A) were contaminants in SANPAH. ANPAH-CLC exhibited absorption maxima at 458,350, and 250 nm with extinction coefficients of 429413,150, and 33,531 M" cm", respectively, in 100% ethanol (Fig. 2). Absorption maxima occurred at 478, 356, and 251 nm with extinction coefficients of 7018, 15,773, and 26,633 M-' cm-l, respectively, in 25 mM sodium phosphate, 1 m~ MgS04, 0.1 m u EGTA, pH 7.2 (data not shown). Reaction of SANPAH with DAC did not change the absorption attributable to the aryl azido moiety nor was the tropolone moiety of colchicine altered (Fig. 3). Values of the absorption peaks measured by IR and the resonance frequencies measured by 'H and 13C NMR confirm the structure of ANPAH-CLC as shown in Equation buffer, pH 7.5, was added. The reaction was monitored by TLC on silica (benzene:methanol, 3:1), and the TLCs were scraped and counted for radioactivity. After 4 h the DAC had been consumed and no further changes in the TLCs were observed. The reaction solution was evaporated to dryness, redissolved in HPLC-grade methanol, and partially purified by thick layer chromatography on a Whatman PLK5F linear K silica gel plate. After elution with HPLC grade methanol, the [3H]ANPAH-CLC was purified to >99% by preparative HPLC on a Beckman CIS reverse-phase column at a flow rate of 5.0 ml/min and a pressure of 1.70 kp.s.i. with a mobile phase of methanokHzO, 9:l. The material eluting at 20.75 ml was collected from eight injections, combined, evaporated to dryness, and reinjected on the preparative HPLC column to check purity (Fig. 1B). The [3H]ANPAH-CLC was dissolved in 4.5 ml of ethanol to give a working stock solution of 6.37 x M. The [3H]ANPAH-CLC had a specific activity of 3.78 Cijmmol.
Binding of ANPAH-CLC to Tubulin-ANPAH-CLC behaved as an apparent competitive inhibitor of the binding of [3H]colchicine based on double-reciprocal graphical analysis of the data (Fig. 4A). However, a secondary plot of the apparent K d for colchicine versus the concentration of AN-PAH-CLC was nonlinear (Fig. 4B). Secondary plots of the reciprocal of the mass of colchicine bound uersus ANPAH-CLC concentration (modified Dixon plot) (24) and the slope of the primary double-reciprocal plot uersus ANPAH-CLC concentration were also curved upward (data not shown). The value of the apparent Kd for colchicine was 0.42 f 0.04 ~L M (mean k S.D., n = 4). The value of the apparent Ki for ANPAH-CLC was 0.28 f 0.03 p~ (mean f S.D., n = 8) in the concentration range of 0.8-1.2 PM ANPAH-CLC. r3H] ANPAH-CLC bound to renal tubulin with the value of the apparent Kd equal to 0.50 f 0.04 pM (mean f S.D., n = 4), as shown in Fig. 5.
Inhibition of Microtubule Formation-Polymerization of renal tubulin at 37 "C was initiated by addition of dimethyl sulfoxide. Microtubule formation, measured by the absorbance change at 350 nm with time, was followed in the absence and presence of various concentrations of colchicine or AN-PAH-CLC. Fig. 6 shows typical data corrected for the back- colchicine concentration. At concentrations in excess of 10 ~L M ANPAH-CLC and 30 p~ colchicine polymerization was inhibited almost completely, but other aggregation and/or precipitation of the tubulin occurred. This was similar to the aggregation effects of vinblastine that occurred at concentrations as low as 1.5 p~ (data not shown). The aggregation effect was observed in the 3 ~L M ANPAH-CLC concentration as evidenced by the absorbance increase occurring during the latter portion of the time profile.
The observed inhibition of tubulin polymerization by 10 p~ colchicine was approximately 50% of the uninhibited tubulin polymerization under the experimental conditions either colchicine or the analog to bind to tubulin. For comparison purposes between colchicine and analog the experimental protocol chosen had the required GTP present before either addition of Me2S0 or MezSO plus analog. Consequently, rapid polymerization begins when the Me2S0 was added whether or not colchicine or the analog was present. This method afforded a simple reproducible way to initiate microtubule formation and provided a way to compare the inhibition caused by colchicine and the analogs. Fig. 7 shows that a 95%'inhibition of microtubule formation was caused by 10 ~L M colchicine when the order of addition of reagents was changed to allow the colchicine to interact with the tubulin prior to initiation of polymerization by addition of MezSO and GTP. The degree of inhibition was increased to 98% with the same colchicine concentration, as shown by the dotted line in Fig. 7, when an incubation time of 15 min was allowed between the colchicine and tubulin and polymerization was initiated by the addition of MezSO followed by GTP. As expected, these results demonstrate that the order of addition and incubation time between the colchicine and tubulin affects the extent of microtubule formation. Similar results are obtained with ANPAH-CLC or other colchicine analogs (data not shown).
Photolabeling of Tubulin with ANPAH-CLC: Spectral Analysis and Controls-Renal tubulin and ANPAH-CLC, at a molar ratio of 1:2, were incubated together, and the UV-VIS spectrum of the solution was measured before and after irradiation. Absorbance at 475 nm decreased after irradiation due to photoreactivity of the aryl azido moiety while the absorbance at 352 nm due to the tropolone moiety was unchanged (Fig. 8). The stoichiometry of photolabeling of tubulin with ANPAH-CLC was calculated from the UV-VIS spectrum of dialyzed photolabeled tubulin (Fig. 9A), and 0.9 mol of AN-PAH-CLC was incorporated per mol of tubulin. This calculation requires the assumption that the molar extinction , """-T""-

Wavelength (nm)
FIG. 8. UV-VIS spectra of a solution of tubulin and 6-(4'azido-2'-nitrophenylamino)hexanoyldeacetylcolchicine be-.fore and after irradiation. ANPAH-CLC and tubulin, at a molar ratio of 2:1, were incubated in sodium phosphate buffer, pH 6.8, for 15 min at about 25 "C. The absorbance was measured as a function of wavelength before (-) and after (---) irradiation for 1 h with a low-pressure Hg lamp as described under "Experimental Procedures." The inset illustrates the loss of absorbance at 475 nm after irradiation. coefficient for the tropolone moiety of ANPAH-CLC is unchanged upon incorporation into tubulin. No absorption peak at 350 nm was detected for tubulin irradiated in the presence of colchicine instead of ANPAH-CLC (Fig. 9B). The results shown in Figs. 8 and 9 demonstrate that colchicine itself is insufficient to photolabel tubulin under these photolysis conditions and that the photoreactivity of ANPAH-CLC is due to the aryl azido moiety. Photolysis of N-succinimidyl-6-(4'azido-2'-nitrophenylamino)hexanoate in the presence of tubulin also did not show any incorporation as measured by UV-VIS absorption spectra (data not shown). Thus, both the colchicine and aryl azido moieties are required for photolabeling of tubulin.
An additional control to show that colchicine does not photolabel tubulin by itself was performed by irradiating [3H] CLC in the presence of tubulin at molar ratios of 1:l and 2:1, respectively, with the low pressure Hg lamp in the absence of the 420-nm cutoff filter. The Hg lamp had a strong emission band at 350 nm. At the initial molar ratios of either 1:l and 21 ([3H]CLC:tubulin) only 2-3% of the radiolabeled colchicine is incorporated into the tubulin. Borisy et al. (25) had previously shown that colchicine bound to tubulin, when irradiated with long wavelength UV radiation, is converted to lumicolchicine which rapidly dissociates from the tubulin allowing the tubulin to polymerize. Our results are in agreement with these observations. Irradiation of a 2:1 molar ratio ([3H]CLC:tubulin) with a high intensity multiple UV lamp system for 1 h showed substantial spectral changes in the colchicine; however, only 9% of the [3H]CLC was incorporated into the tubulin after this extremely vigorous treatment. Clearly, colchicine by itself under any photolysis conditions was incapable of photolabeling tubulin in a specific manner when compared to ANPAH-CLC.
Photolabeling of Tubulin with PHIANPAH-CLC: Radioisotope Analysis-Renal tubulin and [3H]ANPAH-CLC, at several different molar ratios up to 1:2, respectively, were irradiated together and subsequently dialyzed as described under "Experimental Procedures.'' The stoichiometry of incorporation of [3H]ANPAH-CLC into tubulin was calculated from the radioactivity and tubulin mass in the dialyzed photolabeled tubulin. The results are shown in Fig. 10. About 1 mol of ANPAH-CLC was covalently incorporated per mol of tubulin at initial molar ratios of ANPAH-CLC to tubulin up to 2:l. The mol incorporation of [3H]ANPAH-CLC into tubulin was 68 f 4% (x k S.E., n = 12) of the initial moles of the photoaffinity analog. Unincorporated rH]ANPAH-CLC was photodegraded and removed by the dialysis treatment.
Colchicine Binding to Photolabeled Tubulin-Binding experiments with photolabeled tubulin provided further evidence for the specificity of ANPAH-CLC for the colchicinebinding site. Tubulin was photolabeled with ANPAH-CLC at a molar ratio of 1:2 and dialyzed. Control tubulin preparations were obtained by dialysis treatment alone and by photolysis without ANPAH-CLC followed by dialysis. The capacity of each dialyzed tubulin preparation to bind [3H]colchicine was measured as a function of the tubulin mass. The results are shown in Fig. 11. Tubulin that was photolabeled with AN-PAH-CLC had only 8% of the colchicine-binding capacity of the control tubulin preparation. In addition, irradiation of tubulin alone did not significantly affect its colchicine-binding capacity.
Due to the lack of reversibility of the covalent photoincorporation of ANPAH-CLC it was impossible to remove the label and test the tubulin for general protein destruction and/ or denaturation. Present evidence supports the thesis that the protein was not denatured or generally destroyed after photolysis; however, a conformational change in protein structure upon the photoincorporation of ANPAH-CLC could cause the observed reduction of colchicine binding rather than a specific blockage of the colchicine site. The Ki of 0.28 PM for ANPAH-CLC competition of colchicine binding argues against this possibility. The experimental data suggest but do not prove that general protein destruction had not occurred.
Identification of the Photolabeled Subunit-To determine whether photoactivated ANPAH-CLC was covalently incorporated into the a-or &subunit of renal tubulin, the photolabeled tubulin was subjected to gel electrophoresis. Initially the electrophoretic patterns of control and photolabeled tubulin samples were compared. Tubulin was photolabeled with ANPAH-CLC at a molar ratio of 1:2, dialyzed, dissociated in sodium dodecyl sulfate, and subjected to electrophoresis on a 10% acrylamide, 0.8% methylene bisacrylamide slab gel containing 0.1% sodium dodecyl sulfate. Some tubulin samples were reduced and carboxymethylated prior to electrophoresis to maximize separation of the a and P bands. Control tubulin was prepared and analyzed in the same manner except it was not irradiated. Photolabeled and control tubulin samples exhibited the same relative mobilities for the a-and P-subunits and were indistinguishable when subjected to electrophoresis together (data not shown). Thus, photolabeling of tubulin did not alter the mobilities of the a-or @-subunit. Little et al. (26) demonstrated that the a-subunit had a smaller electrophoretic mobility than the P-subunit for renal tubulin.
Tubulin was photolabeled with [3H]ANPAH-CLC at a molar ratio of 1:2, prepared for electrophoresis as described under "Experimental Procedures," and subjected to gel electrophoresis on a 10% acrylamide slab gel containing 0.1% sodium dodecyl sulfate. The gel was stained with Coomassie Blue, destained, and then subjected to fluorography. The electrophoretic patterns of photolabeled tubulin as detected fluorographically and by protein staining are shown in Fig. 12. The [3H]ANPAH-CLC was covalently incorporated into the asubunit of tubulin. There was no significant incorporation into the &subunit even at the largest mass of photolabeled tubulin examined electrophoretically. This specificity of subunit photolabeling was more quantitatively measured by analysis of [3H]ANPAH-CLC-labeled tubulin after electrophoresis on hybrid gels composed of 12.5% acrylamide and 1% ME agarose. Tubulin, photolabeled with [3H]ANPAH-CLC at a molar ratio of 1:1, was subjected to electrophoresis on cylindrical hybrid gels containing 0.1% sodium dodecyl sulfate. The gels were stained with Coomassie Blue and densitometrically scanned at 610 nm (Fig. 13A). A corresponding unstained gel was sliced into 2-mm segments, the segments were melted in H20, and counted for tritium. The results in Fig.  13B show that more than 97% of the radioactivity was present in the a-subunit. Incorporation of rHIANPAH-CLC into the a-subunit of the tubulin was blocked by preincubation of tubulin with colchicine prior to photolysis. Fig. 14 shows that of colchicine and subjected to gel electrophoresis. The gel was sliced into 2-mm segments, melted in water at 80-90 "C, and counted in a liquid scintillation counter. without colchicine significant radioactivity was incorporated into a-tubulin with negligible radioactivity in p-tubulin. HOWever, in the presence of colchicine the level of radioactive incorporation was significantly reduced in a-tubulin.

DISCUSSION
We conclude that the colchicine-binding site is located on the a-subunit of bovine renal tubulin. Binding of the ANPAH-CLC at the colchicine-binding site is based on its apparent competition with colchicine, the loss of more than 90% of the colchicine-binding capacity after photolabeling with ANPAH-CLC, the stoichiometry of incorporation, and protection of the colchicine site on a-tubulin when colchicine is present during photolysis. The one-to-one stoichiometry of photolabeling was obtained with only a 15-min preincubation of ANPAH-CLC with tubulin. Preincubation of ANPAH-CLC with tubulin for 3 h before photolysis (data not shown) yielded the same final stoichiometric results as obtained with the 15min incubation. These results indicate a rapidly reversible binding of the ANPAH-CLC to tubulin. These binding characteristics of ANPAH-CLC -are similar to those previously demonstrated for colchicine by Garland (27) and Lambier and Engelsborghs (28).
Localization of the colchicine-binding site on the a-subunit of bovine renal tubulin is based on analysis of data obtained by gel electrophoresis of [3H]ANPAH-CLC-photolabeled tubulin. This is the first time that the colchicine-binding site of tubulin has been labeled with a specific in situ-activated colchicine derivative. Our results are in agreement with previous observations on mouse brain tubulin using bromocolchicine (6) and porcine brain tubulin using limited proteolysis with [3H]colchi~ine binding (29) that implicate the a-subunit as the colchicine-binding site. In contrast, Ludueiia and Roach found that colchicine and podophyllotoxin blocked the intrachain cross-linking of two sulfhydryl groups in brain tubulin with N,N'-ethylene-bis(iod0acetamide) (30). These sulfhydryl groups correspond to two cysteines present in the &-subunit. They interpreted these results to mean that the colchicine-binding site may be on the p-subunit (31).
Deacetylcolchicine was prepared as the initial reactant because it contains a reactive amino group as well as intact structures of rings A and C of the colchicine molecule (Equation l). The benzenoid ring with at least one methoxy group (ring A) and the troponoid ring with specific positions for the methoxy and carbonyl moieties (ring C) result in the tight binding of CLC to tubulin (32-35). The amino-substituted moiety on ring B and even ring I3 itself are not considered essential (33). The nucleophilic amino group in DAC readily reacts with acids, esters, isothiocyanates, and sulfonyl chlorides for preparation of photoaffinity, spin-labeled, fluorescent, and heme-peptide derivatives of colchicine (8, 36-39).
ANPAH-CLC was synthesized as the photoreactive analog because nucleophilic displacement on the N-succinimidyl moiety of N-succinimidyl-6-(4'-azido-2'-nitrophenylamino) hexanoate by the amino group of DAC yields an amide linkage that very closely resembles the structure of CLC. In addition, the highly reactive nitrene, photochemically generated from the azido group, covalently bonds by insertion into carbonhydrogen bonds (1, 2). Thus, the reactive group is not generated until the analog binds, and the reactive group is not limited to a particular amino acid residue for covalent incorporation. The presence of the nitro group meta to the azido group shifts the wavelength maximum for photolysis to above 400 nm (1, 2, 40) and, thereby, minimizes the possibility of photochemical rearrangements of the troponoid ring of CLC. Retention of the absorption maximum at 350 nm in ANPAH-CLC after irradiation provides a means to measure photoaffinity labeling of tubulin plus indicating that the troponoid ring is not rearranged. The data presented here support the rationale for the synthesis and application of this photoaffinity derivative of colchicine.
The free energies of binding of CLC and ANPAH-CLC to renal tubulin are about the same based on the similarity of similarity of structures. The lack of linearity of secondary plots of the binding data indicates that ANPAH-CLC does not react with tubulin in a simple bimolecular reaction. Several investigators (27, 28, 41) have proposed that the binding of colchicine induces a conformational change in tubulin and that colchicine itself does not interact with tubulin in a simple bimolecular reaction. Nonlinearity of secondary plots of the binding data are a general phenomenon for CLC and colchicine analogs when a sufficiently large concentration range is examined (8,9, 37, 38).
The inhibition of microtubule formation exhibited by the ANPAH-CLC demonstrates its ability to behave in a manner identical to colchicine. It is a slightly more potent inhibitor of microtubule formation than colchicine and at higher concentrations causes aggregation or precipitation that is analogous to the effects caused by vinblastine (42).
Knowledge of the subunit localization of the colchicine binding site on tubulin may be applicable in characterization of mutant forms of tubulin and elucidation of mechanisms of tubulin-ligand interactions per se and with respect to regulation of assembly and disassembly of microtubules. Such studies will be further aided by knowledge of the specific amino acid residues which interact with colchicine. and I R was performed and was c o n s i s t e n t w i t h p u b l i s h e d a n d e x p e c t e d r e s u l t s . A p u r i t y g r e a t e~ t h a n 99% ( b y HPLC a n a l y s i s ) u a s e s t a b l i s h e d .