Enantioselective Total Synthesis of (−)-Zampanolide, a Potent Microtubule-Stabilizing Agent

Abstract

An enantioselective total synthesis of zampanolide has been accomplished using a novel DDQ/Brønsted acid catalyzed cyclization as the key reaction. The synthesis features cross-metathesis to construct the tri-substituted olefin and a ring-closing metathesis to form the macrolactone. The final N-acyl aminal formation was stereoselectively accomplished by an organocatalytic reaction.

The isolation of taxol in the 1970’s and subsequent decade-long studies led to the approval of paclitaxel as an important anticancer drug. ¹ Paclitaxel inhibits cell division by stabilizing microtubule assembly.² Since the discovery of this unique mechanism of action, the search for new agents has become the subject of immense interest. In recent years, epothilones and discodermolide-based antitumor therapeutics have received significant attention. ³ Also, laulimalide and pelorusides were identified as the exciting new generation of microtubule-stabilizing agents.⁴ Both laulimalide and pelorusides have exhibited intriguing synergistic effects with taxol, maintained potent activity against taxol-resistant cell lines and are not substrates for P-glycoprotein-mediated drug efflux pump.^5,6 More recently, Northcote, Miller and co-workers reported that (−)-zampanolide 1 (Figure 1), a 20-membered macrolide also possesses potent microtubule-stabilizing properties.⁷

Figure 1.

(−)-Zampanolide 1 (natural form) and (+)-dactylolide 2 (enantiomer of natural form)

Zampanolide (1) was originally isolated by Tanaka and Higa from the marine sponge Fasciospongia rimosa in Okinawa in 1996.⁸ More recently, (−)-zampanolide 1 was isolated from a Tongan marine sponge, Cacospongia mycofijiensis, and reported to block cell division in the G2/M of the cell cycle.⁷ Uenishi and co-workers reported potent cytotoxic activity of (−)-zampanolide against SKM-1 and U937 cell lines with IC₅₀ values of 1.1 and 2.9 nM respectively.⁹ Furthermore, it is potent against HL-60, 1A9 cells and especially against A2780AD which is quite resistant to paclitaxel.⁷ Zampanolide’s natural abundance is very scarce which has resulted in only limited biological studies. Zampanolide possesses a unique unsaturated macrocyclic structural feature containing only three stereogenic centers and a chiral N-acyl aminal side chain.

The chemistry and biology of zampanolide has attracted much synthetic attention. Smith and co-workers achieved the first total synthesis of (+)-zampanolide, the unnatural antipode, in 2001.¹⁰ Subsequently, both Hoye et al. in 2003¹¹ and Uenishi et al.⁹ in 2009, reported the total synthesis of natural (−)-zampanolide. These syntheses established that the natural (−)-zampanolide core is the opposite enantiomer of the related natural product (+)-dactylolide (2). Dactylolide displayed only modest cytotoxicity. Since its discovery by Riccio and co-workers in 2001 ¹², a number of total syntheses^{13,14,15,16,17,18,19} and a synthetic approach to (+)-dactylolide have been reported.²⁰ It appears that the N-acyl aminal side chain of (−)-zampanolide is important for its potent cytotoxic properties. The key N-acylation reaction typically⁹ provided only 12% yield of zampanolide along with its epimer and bis-acylated product, suggesting that an improvement is necessary for the synthesis of structural variants. Herein, we report an enantioselective synthesis of (−)-zampanolide that can be amenable to the synthesis of N-acyl aminal derivatives.

Our synthetic strategy for (−)-zampanolide (1) is shown in Figure 2. We planned to synthesize the macrocyclic core in a convergent manner in order to prepare a variety of structural analogs. Our strategic bond disconnection of the sensitive N-acyl aminal side chain at C20 provides macrolactone 3, which can be formed from alcohol 4 and acid 5 by esterification followed by a ring-closing metathesis. A similar RCM strategy was first employed by Hoye and Hu.¹¹ The tri-substituted olefin in 4 would be installed by a cross metathesis of 6. The tetrahydropyran ring 6 would be constructed by an oxidative cyclization reaction of a cinnamyl ether derived from β-hydroxy ester 7. The polyene carboxylic acid 5 would be assembled by a Reformatsky reaction with γ-bromo unsaturated ester 8 followed by Wittig olefination of the corresponding aldehyde.

Figure 2.

Synthetic plan for (−)-Zampanolide 1

As shown in Scheme 1, the synthesis commenced with known ester 7, which was readily prepared with excellent enantioselectivity using Noyori hydrogenation as the key step.²¹ Selective protection of the primary alcohol as a TBDPS ether provided 9. Etherification of the secondary alcohol with t-butyl cinnamyl carbonate in the presence of a catalytic amount of Pd(PPh₃)₄ afforded cinnamyl ether 10 in 73% yield. The ethyl ester 10 was converted to allylsilane 11 by employing a modified procedure of Narayanan and Bunnelle to provide 11 in 81% yield.²² Our subsequent plan was to carry out an oxidative Sakurai type cyclization to construct the 4-methylenetetrahydro-2H-pyran ring stereoselectively. For this transformation, we initially explored an oxidative cyclization reaction with DDQ as benzylic/allylic ethers have been converted to carbocation intermediates using DDQ with or without Lewis acids by Mukaiyama and Hayashi,²³ She et al.,²⁴ and Floreancig and co-workers.²⁵ The reaction with DDQ at −40 °C for 3 days resulted in only 30-35% desired product 6 along with unidentified side products possibly due to the presence of the allylsilane functionality. The same reaction in the presence of a variety of Lewis acids also provided similar results. However, oxidative cyclization with DDQ in the presence of a mild Brønsted acid such as PPTS provided the best results. Reaction of 11 with 1.5 equiv of DDQ and 1.5 equiv of PPTS at −38 °C in acetonitrile for 3 h afforded 6 in 81% yield as a single diastereomer (by 1H-NMR analysis). The NOESY data fully corroborated the depicted cis-stereochemistry of 6. Presumably, the cyclization proceeded through a Zimmerman-Traxler transition state where all substituents are equatorially oriented.²⁶

Scheme 1.

Synthesis of tetrahydropyran 14

Our synthetic strategy then called for the elaboration of the E-trisubstituted olefin in 4 by a cross-metathesis reaction. The corresponding olefin substrate 12 was obtained in optically active form by opening of PMB-protected glycidyl derivative with isopropenylmagnesium bromide followed by protection of the resulting alcohol as TES-ether. Our attempted cross-metathesis of 6 with 12 under a variety of conditions did not provide any cross-metathesis product. Assuming the lack of reactivity of the styrene chain, we planned to convert the styrene chain into a methylene chain. This was achieved by selective dihydroxylation of the styrene chain with ADmix-α followed by diol cleavage with NaIO₄ to provide the corresponding aldehyde as described by Smith and Dong.²⁷ Wittig olefination of the resulting aldehyde with methylene phosphorane afforded 13 in 78% yield (3 steps). Cross-metathesis²⁸ of 13 with 12 using Grubbs’ second generation catalyst (10 mol %) in CH₂Cl₂ at reflux for 9 h provided E/Z olefin mixture (1.7:1) in 57% yield. The mixture was treated with HF·py to remove all silyl groups. The resulting alcohols were separated by silica gel chromatography. Photochemical isomerization of the Z-isomer 15 provided a 51% yield (68% brsm) of tri-substituted olefin 14.

The synthesis of polyene carboxylic acid 5 is shown in Scheme 2. Allyl bromide 8 was prepared as an E/Z mixture (1.3:1) using the procedure of Fallis and Lei.²⁹ Reformatsky reaction of the mixture of bromides 8 with acrolein resulted in allylic alcohol 16 (47% yield) and unsaturated δ-lactone 17 (36% yield) after separation by silica gel chromatography. DIBAL-H reduction followed by Wittig reaction with (carbethoxymethylene)triphenylphosphorane afforded allylic alcohol 18. Protection of the resulting alcohol as a PMB-ether followed by saponification of the ester provided acid 5 in 62% yield (2 steps).

Scheme 2.

Synthesis of polyene carboxylic acid 5

The synthesis of macrolactone 3 is shown in Scheme 3. The primary alcohol in 14 was selectively oxidized with TEMPO in the presence of PhI(OAc)₂. The resulting aldehyde was subjected to Wittig olefination with methylenetriphenylphosphorane to provide 19 in 60% yield (2 steps). Esterification of acid 5 with alcohol 19 under Yamaguchi conditions using 2,4,6-trichlorobenzoyl chloride and DMAP furnished ester 20 in 91% yield. Ring-closing metathesis using Grubbs’ second generation catalyst (12 mol %) in benzene at 60 °C for 20 h afforded 3 as a mixture (1:1) of diastereomers at the C7 chiral center. Removal of both PMB-ethers by treatment with DDQ provided the corresponding diol. Dess-Martin oxidation of the diol furnished (−)-2, the unnatural antipode of dactylolide (52% for the 3 steps).

Scheme 3.

The synthesis of (−)-2

The conversion of (−)-2 to (−)-zampanolide 1 required the addition of a carboxamide to the aldehyde to form a N-acyl aminal. Such direct addition in one previous synthesis⁹ afforded only 12% yield of (−)-zampanolide, 12% yield of epimeric product 23 and 16% yield of bis-amide 24. In an effort to improve this reaction, we investigated Brønsted acid-catalyzed N-acyl aminal formation under a variety of reaction conditions. In one of our initial successful attempts, reaction of (−)-2 with carboxamide 21 in the presence of diphenylphoshoric acid clearly provided a mixture of all three products, (−)-zampanolide, epi-zampanolide, and bis-amide 24, although analysis of the mixture showed that (−)-zampanolide formation was favored slightly. Encouraged by this result, we then investigated N-acyl aminal formation between aldehyde (−)-2 and amide 21 in the presence of matched chiral phosphoric acid (S)-TRIP^30,31 (22, 20 mol %) at 23 °C for 12 h. These reaction conditions provided excellent chemoselectivity towards mono addition and the formation of bis-amide 24 was not observed. The reaction furnished (−)-zampanolide in 51% yield and epi-zampanolide 23 in 18% yield after separation/purification by HPLC. We have also investigated the corresponding mismatched reaction with (R)-TRIP, however, this reaction afforded a 1:1 mixture of (−)-1 and epi-1. Again, the formation of bis-amide 24 was not observed. The spectral data (¹H NMR and ¹³C NMR) of synthetic 1 ([α]²²_D - 94, c 0.08, CH₂Cl₂) is in agreement with that of natural zampanolide (Lit.⁸ [α]²²_D - 101, c 0.12, CH₂Cl₂).

In summary, we have accomplished an enantioselective synthesis of (−)-zampanolide. The synthesis features a novel intramolecular oxidative cyclization reaction, a cross-metathesis reaction to construct a tri-substituted olefin, a ring-closing metathesis to form a highly functionalized macrolactone, and a chiral phosphoric acid catalyzed stereoselective N-acyl aminal formation that stereoselectively furnished (−)-zampanolide and no bis-amide byproduct. The present synthesis is amenable to the synthesis of structural variants of (−)-zampanolide and further investigations of important analogs are in progress.

Scheme 4.

The synthesis of (−)-Zampanolide