|Home | About | Journals | Submit | Contact Us | Français|
The total synthesis of 22-(3-azidobenzoyloxy)methyl epothilone C is described as a potential photoaffinity probe to elucidate the β-tubulin binding site. A sequential Suzuki-aldol-Yamaguchi macrolactonization strategy was utilized employing a novel derivatized C1–C6 fragment. The C22-functionalized analog exhibited good activity in microtubule assembly assays, but cytotoxicity was significantly reduced. Molecular modeling simulations indicated that excessive steric bulk in the C22 position is accommodated by the large hydrophobic pocket of the binding site. Photoaffinity labeling studies were inconclusive suggesting non-specific labeling.
The epothilone class of natural products, typified by epothilone A (1a), epothilone B (1b), and epothilone C (2), is an extremely cytotoxic family of macrocyclic compounds. Their isolation from the myxobacterium Sorangium cellulosum was first reported in 1992 by Gerth et al.1 It is well established that the binding sites on β-tubulin for the epothilones and taxanes overlap. However, early observations that the epothilones maintained toxicity against paclitaxel resistant cell lines, in which the resistance arose from tubulin point mutations in the paclitaxel binding site, suggested that the mode of binding was significantly different.2 As a result, early attempts to find a shared pharmacophore between paclitaxel and epothilone B were unsuccessful.3–6 The clinical importance7 of this class of compounds has led to an explosion of synthetic activity8 –10 and a truly multi-disciplinary effort has culminated in two β-tubulin-epothilone binding models. The first model is based on combined molecular modeling, NMR, and electron-crystallography (EC) data of Zn2+ induced tubulin sheets,11 while the second model is derived from NMR studies of epothilone bound to unpolymerized α, β-tubulin dimer in solution.12
In the EC-model the C7 hydroxyl, the C5 carbonyl, and the C3 hydroxyl of epothilone engage with Arg282 and Thr274 via H-bonds, while the C1 carbonyl interacts with Arg276. The thiazole ring then forms a hydrogen bond with His227 in the H7 helix. Although they are remote from the binding site, the residues Gln292 and Ala231 are important in maintaining the correct conformation of the M-loop and H7 helix, respectively.
The NMR-model supports the interaction between the C7 hydroxyl and Arg282 and Thr274 and further, that this interaction locks Arg276 into position to form a salt bridge with Asp224. The thiazole and His227 engage in a π–π interaction and are not hydrogen bonded. Finally, and perhaps most intriguingly, is the involvement of the hydrophobic pocket hosting Phe270, which is vital for paclitaxel activity and also hosts the methyl group of epothilone B and thus accounts for its 10-fold increase in activity over epothilone A.
The underlying consensus of both the EC-model and NMR-model is a rigidification and alignment of the M-loop and the H7 helix, which leads to a conformational change in the α, β-tubulin dimer. This hypothesis is in agreement with a different overall macromolecular structure observed between polymerized and unpolymerized tubulin.13
Mammalian brain tubulin, the most abundant and studied of the tubulins, is highly heterogeneous due to the existence of seven α- and eight β-isoforms and several types of post-translational modifications. These combinations are responsible for producing greater than 22 bands when mammalian brain tubulin is separated on high-resolution isoelectric focusing gels.14 Given this complexity there still remains significant advances to be made in the understanding of microtubule stabilizing agents. Early information on the paclitaxel binding site was derived through photoaffinity labeling.15–20 More recently discodermolide photoaffinity labeling studies have been reported.21 However, to date the photolabeling with epothilone derivatives has remained elusive. Toward this end we have synthesized the C22-photoreactive derivative 3. We now report the synthesis of 3 and the resulting microtubule assembly, cytotoxicity studies, and photoaffinity labeling studies.
The synthesis of the C22 analog follows the well precedented aldol/macrolactonization approach.22 With the known aldehyde 12 in hand we focused our attention on the synthesis of the modified C1–C7 fragment 11 (Scheme 1). Starting from the known β-hydroxy ester 4 two sequential Frater alkylations where carried out. Thus, the β-hydroxy ester 4 was treated with 2 equiv of LDA and the resulting dianion quenched with 1.1 equiv of allyl iodide to give the anti-allylated compound 5 in 82% yield. The procedure was subsequently repeated and quenched with MeI to provide the desired anti-methylated product 6 in 66% yield. The free alcohol was protected as the TBS ether (94%) and the alkene function of 6 was ozonolyzed to furnish aldehyde 7 (97%). The aldehyde was subsequently reduced and the resulting alcohol protected as the trityl ether.
The C6–C7 ethyl chain was then introduced by first converting the ester 8 to aldehyde 9 through a two-step reduction (80%)/oxidation (89%) sequence. The aldehyde 9 was then treated with EtMgBr and the resulting alcohol (71%) was oxidized to provide ketone 10 in 89% yield. Finally, the benzyl group was removed and the free primary C1 alcohol reprotected as TBS ether 11 in preparation for the aldol reaction.23
Accordingly, the C1–C7 fragment 11 was treated with LDA and then the aldehyde fragment 12 was added to give adduct 13 in 38% yield (Scheme 2). The newly formed alcohol function was protected as the TBS ether (90%) and the C1 primary TBS ether was removed (72%) subsequently. The resulting alcohol was sequentially oxidized to the aldehyde (98%) and then to the acid 14 (94%).
The TBS-protecting group on the allylic alcohol 14 was deprotected (96%) and the macrocycle closed by employing a Yamaguchi macrolactonization to form the core skeleton 15 in 77% yield. The trityl group was removed with CSA to give the corresponding primary hydroxyl derivative in 50% yield. Finally, the meta-azidobenzoic acid was converted to the acid chloride on treatment with oxalyl chloride, which was subsequently reacted with the epothilone derivative to form photoaffinity adduct 16 in 82%.24 Removal of the TBS groups with TFA then gave the target compound 3 (Scheme 2).
The microtubule assembly properties and cytotoxicity of the meta-azido analog 3 are summarized in Table 1. In the microtubule assembly assay the photoaffinity analog 3 was found to assemble microtubules in a concentration-dependent fashion.25 However, the assembly was not as rapid and did not reach the same extent as in the presence of epothilone B. Comparison of the ED50 values shows that 3 is about 20% as effective as epothilone B. Compound 3 is much less potent against NCI-ADR and MCF-7 cancer cell lines. The very low cytotoxicity could be due to poor uptake of 3 by the cells.
The labeling studies, carried out on assembled microtubules with 3H-meta-azido analog 3, demonstrated that irradiation led to incorporation of label into the protein. However, competition studies showed that although 40 μM epothilone B did not reduce the labeling by 10 μM analog 3, 40 μM paclitaxel was able to displace approximately 50%. These data suggest that, although analog 3 can promote the assembly of microtubules, it might bind in such a way that it cannot be displaced by epothilone B or fully by paclitaxel. Another possibility is that 3 is displaced but when it is free in solution and irradiated, it produces reactive intermediates that react in a non-specific manner with the protein.
Given that we observed microtubule assembly with 3, and that up to 50% of the binding was inhibited with Taxol, docking studies were conducted to probe tubulin binding modes and ligand–receptor interactions for analog 3. Figure 1a shows the docked conformation of 3 in the β-tubulin binding site.11 Figure 1b shows a comparison of the docked configurations of epothilone B (yellow) and analog 3 (cyan) in the β-tubulin active site. In analog 3, the bulky 22-(3-azidobenzoyloxy)methyl group slightly reorients the hydrophobic C8–C15 portion of the compound to avoid unfavorable van der Waals interactions with the bulky C22 position. This relatively minor rearrangement does not significantly disrupt the hydrogen bonding network between receptor and the C1–C7 portion of the ligand. However, in analog 3, the thiazole hydrogen binds to Arg276 rather than to His 227 as is found with epothilone B.
In summary, the photoaffinity analog 3 promotes the assembly of tubulin into microtubules and the computer modeling studies revealed key ligand-receptor interactions. Specifically, molecular modeling simulations indicated that excessive steric bulk in the C22 position is adequately accommodated by the large hydrophobic pocket in the binding site. However, the non-specific covalent labeling limits the practical utility of this probe for further elucidation of the β-tubulin binding site.
This work was supported by the National Institute of Health grants CA79641 and CA105305. E.A.R. and J.F.G. were supported by NIH Training Grant GM 07775. E.A.R. was also supported by DAMD 17-001-0303.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errorsmaybe discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.