Toward More “Ideal” Polyketide Natural Product Synthesis: A Step-Economical Synthesis of Zincophorin Methyl Ester

Abstract

A highly efficient and step-economical synthesis of zincophorin methyl ester has been achieved. The unprecedented step-economy of this zincophorin synthesis is principally due to an application of the tandem silylformylation-crotylsilylation/Tamao oxidation-diastereoselective tautomerization reaction that achieves in a single step what would typically require a significant multi-step sequence.

Polyketide natural products continue to influence small molecule drug development efforts. Both natural products (e.g. discodermolide¹) and designed analogs thereof (e.g. fludelone²) have progressed into clinical trials, and it seems reasonable to anticipate that it might only be a matter of time before approved drugs begin to emerge from such medicinal chemistry programs. It is equally reasonable to anticipate that most such compounds will have to be synthesized (as will, of course, most analogs), as was certainly the case for both discodermolide³ and fludelone. It is for this reason that, despite decades of beautiful, powerful, and profoundly influential chemistry devoted to the synthesis of such structures, there remains a great need for creative new approaches that achieve significantly greater levels of “ideality.”⁴ Progress in this regard would be expected to have an impact on every aspect of polyketide natural product-based synthesis and drug development efforts.

Zincophorin (1) and its methyl ester (2)⁵ have been popular targets for synthetic chemists ever since the groundbreaking synthesis by Danishefsky in 1987 (Figure 1).^6,7 Two additional total syntheses have been reported since then by Meyer and Cossy⁸ and by Miyashita⁹ (in addition to numerous reports of fragment syntheses¹⁰), and, interestingly, all three syntheses (of 2) required ~47/~52 total steps¹¹ with longest linear sequences of 36, 28, and 37 steps resepectively. As part of a broad program devoted to the development of highly efficient and step-economical syntheses of polyketide structures of this type,¹² we decided to undertake a new synthesis of zincophorin methyl ester. Our primary motivation was to set for ourselves the goal of completing the synthesis in about half the number of total steps as the three previous syntheses, because we felt that achieving this would require a fresh approach and true methodological innovation (i.e. greater ideality). We report here the results of these efforts that have culminated in a 27/31 total step synthesis of 2.

Figure 1.

Zincophorin and zincophorin methyl ester.

The synthesis commenced with an asymmetric epoxidation of alkene 3¹³ using Shi’s catalyst¹⁴ to provide 4 in 87% yield and 90% ee (Scheme 1). Epoxide opening using Pagenkopf’s procedure¹⁵ gave 5 in 43% yield.¹⁶ NaH-catalyzed silane alcoholysis with di-cis-crotylsilane^12d then provided 6 in 97% yield and set the stage for an application of the tandem silylformylation-crotylsilylation/Tamao oxidation-diastereoselective tautomerization reaction.^12k Applied to 6, this complex series of chemical events produced 7 in 67% yield with ≥15:1 overall diastereoselectivity. The transformation of 6 into 7 (which we have carried out on multi-gram scale) is remarkable not only for the direct installation of a ketone, three stereocenters, and an alkene, but also for the simplicity of the starting materials (a crotyl-SiH fragment, a propynyl fragment, CO, and H₂O₂). Overall, 7, which contains five of the ten stereocenters of the C(1)–C(16) fragment, is accessed in just four steps from 3, and this sequence is further noteworthy for what is not employed: protecting groups, non-strategic redox reactions, and chiral auxiliaries, controllers, and/or reagents. Using Baran’s algorithm, this adds up to a four step sequence that delivers five stereocenters with 100% ideality.⁴

Scheme 1.

A four-step synthesis of 7 from 3.

A series of straightforward steps (selective protection to give 8, syn-selective β-hydroxy ketone reduction¹⁷ to give 9, diol protection to give 10, and alkene oxidative cleavage) converted ketone 7 into aldehyde 11 and set up a crotylation reaction to establish the C(6) and C(7) stereocenters (Scheme 2). The desired product, 12, is the result of Felkin addition of a Type I trans-crotylmetal reagent to aldehyde 11, and this is a fully matched case¹⁸ that should not require external asymmetric induction for very high levels of diastereoselectivity.¹⁹ A survey of various Type I trans-crotylmetal reagents revealed that the potassium trans-crotyltrifluoroborate reagent introduced by Batey²⁰ was possessed of superior characteristics from the perspective of both efficiency and practicality/ease of use. In the present case its use led to the isolation of 12 in 85% yield (from 10) with ≥20:1 diastereoselectivity.

Scheme 2.

Synthesis of the C(11), C(7), and C(6) stereocenters.

The final stereochemical challenges in the synthesis of the C(1)–C(16) fragment were the C(2) and C(3) stereocenters that would accompany tetrahydropyran ring synthesis. It was clear that the most direct way to accomplish those goals in a single step would be the addition of a propionate enolate to an oxocarbenium ion at C(3). To set up such a reaction, 12 was subjected to hydroformylation to give hemiacetal 13 which was acetylated to give 14 in 94% overall yield (Scheme 3). While the well-established preference for axial attack on the oxocarbenium ion generated from 14 would give the desired outcome at C(3), control of the C(2) center was much more speculative. Extensive experimentation with various achiral propionate enolate species failed to reveal an adequate solution, and we therefore turned to the use of chiral enolates that would allow for control of enolate face selectivity. Romea and Urpí have developed a protocol for the highly stereoselective addition of the titanium enolate derived from 15 to oxocarbenium ions derived from acetals, glycals, and pseudoglycals,²¹ and this appeared to be a highly relevant precedent. Indeed, the titanium enolate derived from 15 was treated with 14 and SnCl₄ to produce 16 in 91% yield as a single diastereomer. Methanolysis proceeded exceptionally smoothly to give 17 and this was followed by a three-step conversion of the benzyl ether into the N-phenyltetrazolylsulfone 20.

Scheme 3.

Completion of the C(1)–C(16) fragment.

The synthesis of the (17)-C(25) fragment commenced with a Sc(OTf)₃-catalyzed crotylation of propionaldehyde using cis-crotylsilane 21²² (Scheme 4).²³ This reaction proceeded smoothly at ambient temperature to provide 22 in 97% yield (based on the use of 21 as the limiting reagent) and 93% ee. Highly trans-selective (>20:1) cross-metathesis with excess methacrolein and the second generation Hoveyda-Grubbs catalyst²⁴ was followed without purification by alcohol tosylation using the Tanabe protocol²⁵ to provide 23 in 79% yield. A second application of the Sc(OTf)₃-catalyzed crotylation reaction with trans-crotylsilane 24 then gave 25 in 81% yield with excellent (19:1) diastereoselectivity. Protection of the alcohol as its para-methoxy benzyl (PMB) ether was followed in the same pot by tosylate reduction with LiBEt₃H to give 26 in 86% yield. Finally, one pot oxidative cleavage produced aldehyde 27 in 87% yield. The synthesis of 27 thus proceeded in just five steps and 46% overall yield from 21 and relied on two applications of the operationally attractive Sc(OTf)₃-catalyzed crotylation methodology.

Scheme 4.

Synthesis of the C(17)–C(25) fragment.

Julia-Kociensky olefination²⁶ with sulfone 20 and aldehyde 27 proceeded smoothly and with excellent trans selectivity to provide 28 in 69% yield (Scheme 5).²⁷ Three sequential deprotection steps (oxidative PMB removal, basic carbonate methanolysis, and TBS deprotection) then completed the synthesis of zincophorin methyl ester 2 in 60% overall yield. Full spectral comparison to the data provided by Cossy^8b and Miyashita⁹ confirmed the identity of our synthetic material.

Scheme 5.

Completion of the synthesis.

This synthesis of zincophorin methyl ester proceeds in 27/31 total steps,¹¹ with a longest linear sequence of 22 steps from (E)-4-hexen-1-ol in 4.2% overall yield.²⁸ Another useful measure of efficiency is steps/stereocenter,²⁹ and in this regard it is noteworthy that the synthesis of the C(1)–C(16) fragment 20 – which contains 10 stereocenters – required just 1.8 steps/stereocenter. Regardless of the metrics used to guage efficiency, it is clear that much of the effciency and step-economy of the route derives from the four-step transformation of 3 to 7. The “ideality” of that sequence is without precedent and we remain committed to the further development of these and related transformations for application to the synthesis of important and complex polyketide natural products and analogs thereof.