A Formal Synthesis of (−)-Englerin A by RRCM and Transannular Etherification

Abstract

A bicyclization approach to englerin A has culminated in a formal asymmetric total synthesis. Key transformations in the 10-step sequence are a regiospecific epoxide opening and a relay ene-yne-ene metathesis that converts linear substrates specifically to Δ^4,6-guaiadiene-9,10 diol derivatives. Regiospecific functionalization of the diene moiety installs the oxygen bridge required for the englerin tricyclic core.

(−)-Englerin A is a natural product from Phyllathus engleri, a plant common in East Africa.¹ It displays selective and potent inhibition of the growth of renal cancer cell lines in the NCI-60 screen.

The compact and intriguing structure of this new lead has inspired several total syntheses.² The strategies conceived for the total synthesis of (−)-englerin A have provided analogs and these have contributed to the development of a preliminary structure-activity relationship (SAR).³ In light of the unique activity profile of (−)-englerin A, one can anticipate optimization of its pharmacological properties by the exploitation of mechanism of action studies and also by more extensive medicinal chemistry. In this context, each total synthesis is valuable in that it establishes proof-of-principle for a synthetic route to the tricyclic framework and also provides opportunities for variations on the scaffold itself.

Our analysis of the structure of the englerin core (Scheme 1) led us to consider the hydroazulene 3 as the product of an ene-yne-ene metathesis-based bicyclization.⁴ In this approach, bicyclic 3 would arise from tandem ring closure initiated by the ruthenium carbene 4. Site-specific generation of this reactive intermediate required that our strategy be developed to include a relay metathesis step.⁵ Thus, we designed substrate 5. A metathesis cascade, specifically and conveniently initiated at the monosubstituted olefin of allyl ether 5, was envisioned to proceed by way of ruthenium carbene 4 to give the desired 3.

Scheme 1.

Retrosynthetic Analysis

In order to complete a synthesis based on hydroazulene 3 with efficiency, we would have to differentiate the two double bonds so as to further the synthetic scheme. We postulated that reversible addition of a soft electrophile to diene 3 would be accompanied by transannular etherification across the 7-membered ring, leaving the cyclopentene olefin untouched (3 -> 2). Indeed, examination of the two unstrained conformations available to an intermediate bridged cation 6 (Figure 2) led us to believe that the hydroxyl nucleophile would effect ring opening by attack at C-7 (conformation 2) as it cannot adopt a position that would allow backside attack at C-6.

Figure 2.

Two conformations available to bridged cation 6

Our strategy was attractive because it appeared that the relay ring closing metathesis (RRCM) substrate 5 would be readily available from geraniol (7a). Opening the ring of a 2,3-disubstituted epoxy alcohol at the 3-position (see 9 -> 10) seemed likely to provide access to the desired functionality pattern for substrate 5. Implementation of a scheme based on these original concepts has now progressed to a formal total synthesis of (−)-englerin A.

Geraniol (7a) was converted to a mixture of diol derivatives 5b and 13b in seven steps (Scheme 2). O-Allylation followed by catalytic SeO₂ oxidation⁶ gave a useful yield of the (E)-allylic alcohol 8. Then Sharpless epoxidation⁷ introduced chirality⁸ and provided an opportunity to incorporate the required alkyne substituent with the desired stereochemistry at the latent C-1 (see 9).

Scheme 2.

Synthesis of RRCM Substrate 5

We were surprised that we were unable to find an unbiased example of an acetylide opening of a 2-alkyl 2,3-epoxy alcohol or a derivative of such an alcohol.^9,10 Consequently we tested conditions reported to effect ring opening of 3-substituted 2,3-epoxy alcohols and their derivatives. Of these, the most successful was treatment of the epoxide opening with Li-acetylide ethylene diamine complex in HMPA/DMSO.¹¹ This protocol provided diol 10 in 82% yield.

Parikh-Doering oxidation¹² gave aldehyde 11 which was subjected to a Barbier addition¹³ with the reagent from 2-bromomethyl-3-methyl-1-butene (12).¹⁴ This produced a mixture of the desired diol 5a and its C-9 epimer, diol 13a. With the goal of obtaining derivatives that would allow us to assign unambiguously the relative stereochemistries of the two diastereomers and to test metathesis conditions, we prepared the cyclic carbonates 5b and 13b by treating the diol mixture with carbonyl diimidazole (CDI).

Relay ring closing metathesis (RRCM) of the carbonate mixture (5b+13b) with the Stewart-Grubbs catalyst¹⁵ gave a mixture of stereoisomers 14 and 15 (Scheme 3). This reaction is impressive in several respects: (1) only the guaiadiene ring system is generated; no decalin derivative was detected,¹⁶ (2) the formation of the cyclopentene ring illustrates the capability of relay-initiated enyne metathesis to produce tetrasubstituted olefins, and (3) both carbonate stereoisomers 5b and 13b undergo bicyclization, giving 14 and the more conformationally rigid 15, indicating the power of the Stewart-Grubbs catalyst.

Scheme 3.

RRCM and Preparation of Transannular Etherification Substrate

The major compound, isolated by chromatography in 45% yield, was assigned structure 14 on the basis of nuclear Overhauser effects (see the Supporting Information). Hydrolytic removal of the carbonate group gave diol 3 (R = H) and selective silylation of the secondary alcohol gave cyclization substrate 3 (R = TBS) in 90% yield for the two steps.

After some experimentation, we found conditions for introduction of the oxygen bridge. Our hope of effecting transannular opening of a Δ^6,7 epoxide was disappointed when treatment of diene 3 (R = TBS) with mCPBA provided a product in which the olefinic proton was retained. Attempts to connect the oxygen bridge by haloetherification were uniformly unsuccessful, leading in most cases to the recovery of starting material. Oxymercuration with Hg(O₂CCF₃)₂ followed by addition of NaCl/NaHCO₃ solution¹⁷ gave the alkyl mercurial 16 from regio- and stereospecific addition to the Δ^6,7 olefin¹⁸ (Scheme 4). Therefore, product formation was consistent with the proposed concerted addition portrayed in Figure 2.

Scheme 4.

Oxymercuration / oxidative demercuration of hydroazulene 3

Oxidative demercuration¹⁹ with NaBH₄ and O₂ in DMF provided a mixture of stereoisomeric tertiary alcohols 17 and 18 in which the functionality pattern of the allylic system has been reversed. Separation of the two diastereomers resulted in isolation of a 55% yield of the α-alcohol 17 and 37% yield of the β-isomer 18. Although the ionic transannular oxymercuration reaction was specific in the desired sense, oxidative demercuration through the radical intermediate resulted in oxygenation at C-4 rather than at C-6.

Although we would have preferred to obtain the known Δ^4,5 C-6 alcohol 19²⁰ directly, alcohol 17 has been converted to (−)-englerin A in seven steps (by way of alcohol 19).²⁰ Therefore, access to alcohol 17 completes a formal synthesis of (−)-englerin A (1).

The synthesis of alcohol 17 illustrates the efficient opening of the epoxide ring of a β-substituted α-epoxy alcohol under the lithium acetylide conditions and the relay ene-yne-ene metathesis method for the preparation of a bicyclic diene that is disubstituted on both ends and that contains a tetrasubstituted olefin. Furthermore, the conversion of diene 3 to the mercurial 16 provides proof-of-concept for the soft electrophile-initiated regio- and stereoselective transannular etherification of Δ^4,5, Δ^6,7 guaiadienes. Approaches to compounds in the englerin series remain under investigation in our laboratories.

Figure 1.

Structure of englerin A