Jillian E.
Spangler
,
Cheryl A.
Carson
and
Erik J.
Sorensen
*
Department of Chemistry, Princeton University, Frick Chemical Laboratory, Princeton, NJ 08544, USA. E-mail: ejs@princeton.edu
First published on 11th June 2010
A stereodivergent synthesis of the [3.3.1] bicyclic core of edaxadiene was completed utilizing a key intramolecular oxidative ketone allylation. Significant discrepancies between the spectroscopic data obtained for the synthetic construct and the natural isolate raised questions about the structural assignment of edaxadiene. A subsequent structural reassignment was validated by completion of a total synthesis of the correct structure of the natural product.
Our group has a growing interest in molecules related to the biology and treatment of tuberculosis.5 The biological importance as well as the scarcity of edaxadiene (1) led us to actively pursue a total synthesis of this intriguing natural product. Our goal was to produce sufficient quantities of this molecule to confirm its structural assignment and enable a search for its biomolecular target.
We recognized that at the heart of this challenge is the synthesis of a [3.3.1] bicyclic core bearing five contiguous stereogenic centers, including two all-carbon quaternary stereogenic centers. In addition, the relative configuration of the C13 stereogenic center was not assigned during structural elucidation. Although edaxadiene (1) has conservation of the halimane skeleton (Fig. 1), formation of a unique C7–C13 bond locks the conformation of the B-ring and produces a highly unfavorable 1,3-diaxial interaction between the C17 methyl group and C1 methylene that is absent in related diterpenes.6
Fig. 1 Proposed structure of edaxadiene. Stereochemistry is undefined at the starred (*) carbon. |
We envisioned that edaxadiene could arise from enone 3 by annulation of the leftmost ring onto the convex face of the [3.3.1] bicycle with subsequent deoxygenation of the C6 ketone (4) to form the C5–C6 olefin (Scheme 1). We reasoned that this enone (3) could be fashioned through an intramolecular ketone allylation of an enone of type 2 to forge the C7–C13 bond. Importantly, we anticipated that variation of olefin geometry and reaction conditions in this cyclization would provide access to both epimers at the unassigned C13 stereogenic center.
Scheme 1 Proposed synthesis of 1via an intramolecular ketone allylation. Stereochemistry is undefined at the starred (*) carbon. |
Our efforts thus began with the development of a synthesis of enone 2. The thermal Diels–Alder cycloaddition of tiglaldehyde and a modified Rawal diene7 (5) provided silyl enol ether 6 (Scheme 2). Subsequent methylenation of the aldehyde provided alkene 7 in good yield on a large scale (>15 g). Regioselective hydroboration of the terminal olefin with 9-BBN furnished a B-alkyl borane that was subjected to a Suzuki–Miyaura cross-coupling reaction8 with (Z)-2-bromo-2-butene (10a) or (E)-2-bromo-2-butene (10b). The geometrical isomers 9a and 9b were obtained after deprotection of the enol ether and elimination of oxazolidinone with tetra-n-butylammonium fluoride.
Scheme 2 Diels–Alder-mediated synthesis of enones 9a,b. Conditions: (a) tiglaldehyde, toluene, 120 °C, 80%; (b) Ph3PCH3Br, n-BuLi, THF, 0 °C, 67%; (c) 8a: 9-BBN, THF, reflux; then 10 mol% PdCl2(dppf), aq. K3PO4, 10a, DMF, 50 °C; 8b: 9-BBN, THF, reflux; then 10 mol% PdCl2(dppf), aq. K3PO4, 10b, DMF, 50 °C; (d) 9a: TBAF, THF, 86% (2 steps); 9b: TBAF, THF, 100% (2 steps). 9-BBN = 9-borabicyclo(3.3.1)nonane; dppf = 1,1′-bis(diphenylphosphino)ferrocene; TBAF = tetra-n-butylammonium fluoride. |
After exploring multiple methods for constructing the C7–C13 bond to provide a [3.3.1] bicycle we found that oxidative cyclization of 9a with a bimetallic system of manganese(III) acetate and copper(II) acetate in a mixture of acetic acid and benzene (3:1) provided the volatile [3.3.1] bicycle 12a in moderate yield as a single stereoisomer (Scheme 3).9 This cyclization most likely proceeds through the intermediacy of an A(1,3) minimized intermediate (11a) to provide 12a. Utilizing the same reaction conditions enone 9b underwent cyclization to provide 12b, the opposite C13 epimer, albeit with a lower degree of diastereoselectivity. We were pleased to find that we could gain access to both C13 epimers by controlling the geometry of the starting trisubstituted olefin.
Scheme 3 Synthesis of the [3.3.1] bicyclic core of 1via a stereodivergent intramolecular oxidative ketone allylation. Conditions: (a) Mn(OAc)3·2H2O, Cu(OAc)2·H2O, 80–100 °C, AcOH–benzene (1:3), 12a: 57%, 12b: 51% (1.4:1 dr); (b) NaBH4, MeOH; (c) LiAlH4, Et2O, 0 °C, 13a: 87% (2 steps), 13b: 73% (2 steps); (d) Martin sulfurane, benzene, 80 °C, 14a: 23%, 14b: 34%. |
It was at this point in our efforts that we began to note discrepancies between the reported 13C shifts of the [3.3.1] core of edaxadiene and our synthetic constructs. To address our scepticism of the proposed structure, we converted the diastereomeric bicycles 12a,b into dienes 14a,b, bearing the full [3.3.1] bicyclic core of structure 1. Thus, stepwise reduction of each enone to the saturated alcohol (13a and 13b) was followed by dehydration to the corresponding alkene (14a and 14b) with Martin sulfurane in refluxing benzene.10 The unusually high temperature requirement for the dehydration is attributed to the equatorial disposition of the substrate alcohols. Although the high volatility and non-polarity of these compounds complicated their isolation, we were able to obtain sufficient quantities of each C13 epimer for characterization via semi-preparative gas chromatography. The spectroscopic data for each C13 epimer differed dramatically from the reported spectra of edaxadiene; most notably, the reported 13C-NMR chemical shift at C13 was substantially different (Δδ of >30 ppm) from the 13C chemical shifts observed for bicycles 14a,b (Fig. 2).
Fig. 2 Discrepancies in 13C-NMR shifts that indicate the structure of edaxadiene has been incorrectly assigned. |
These significant differences between the 13C-NMR data of the isolated 1 and our synthetic bicycles (14a and 14b) led to our re-evaluation of the reported spectroscopic data for 1. Our consideration of this data led us to propose compound 15 (Fig. 3) as the actual structure of edaxadiene. This proposal is consistent with the 13C-NMR data, most notably the deshielded nature of C13. Observation of the parent [M − H2O]+ ion by EI mass spectrometry is presumably due to the facile fragmentation of the tertiary allylic alcohol at C13 of 15. We further suspect that edaxadiene, as reported, is likely not a pure compound, but rather a diastereoisomeric mixture at the remote C13 stereogenic center, as indicated by fine doubling of the olefinic protons in the 1H spectrum of 1. Our confidence in this reassignment was bolstered by a comparison of this structure to known terpenes in the literature, which indicated that this was a previously identified halimane diterpene, nosyberkol (15), isolated in 2004 from extracts of the Red Sea sponge Raspailia sp.11 Furthermore nosyberkol (15, also referred to as isotuberculosinol) was previously speculated to be the product of the Rv3378c enzyme produced by M. tuberculosis.12
Fig. 3 Structures of nosyberkol and tuberculosinol. Stereochemistry is unassigned at the starred (*) carbon. |
This proposed structural reassignment led us to develop a synthesis of nosyberkol (15) for structural verification.13 Our efforts began with an exo-selective Diels–Alder cycloaddition14 of diene 17 (available in two steps from 2,2-dimethylcyclohexanone)15 and ethyl tiglate to give the desired cycloadduct as an inseparable mixture of diastereoisomers (2:1) (Scheme 4). Subsequent reduction of the esters and separation of the primary alcohols provided the desired C10 epimer.16 Aldehyde 18 was obtained in good yield using the conditions of Parikh and Doering.17 Our attempts to perform a homologation of 18 through reaction with a phosphonate or phosphonium ylide led to no product formation; however, an aldol condensation with the sodium enolate of acetone and conjugate reduction of the resultant enone with Wilkinson's catalyst18 allowed the desired homologation to ketone 20. Our synthesis of nosyberkol was completed by the addition of vinylmagnesium bromide to the ketone to provide the desired compound as a 1.5:1 diastereomeric mixture at the tertiary allylic alcohol-bearing carbon. The spectroscopic data obtained for synthetic nosyberkol (15) were identical with those reported for both natural nosyberkol and edaxadiene (Fig. 2, see ESI†).19
Scheme 4 Syntheses of nosyberkol and tuberculosinol via an exo-selective Diels–Alder reaction. Conditions: (a) ethyl tiglate, neat, 160 °C, 71% (2:1 exo:endo); (b) LiAlH4, THF, 40 °C, 56% (+24% endo isomer); (c) SO3·pyridine, NEt3, CH2Cl2–DMSO, 0 °C, 86%; (d) acetone, NaHMDS, THF, −78 → 23 °C, 87%; (e) 10 mol% Rh(PPh3)3Cl, HSiEt3, CH2Cl2, 40 °C, 83%; (f) vinylmagnesium bromide, THF, 0 °C, 93% (1.5:1 dr at the starred (*) carbon); (g) Ph3PCH3Br, KHMDS, 0 °C, THF, 91%; (h) 9-BBN, THF, 80 °C; then 10 mol% PdCl2(dppf), Ph3As, Cs2CO3, 21, DMF, 73%; (i) 20 mol% CuCl2, acetone, 20% (dr = 1:1 at the starred (*) carbon). 9-BBN = 9-borabicyclo(3.3.1)nonane; dppf = 1,1′-bis(diphenylphosphino)ferrocene. |
We were also intrigued by the reported biomimetic conversion of tuberculosinol20 (16) (Fig. 3) into edaxadiene (15) by treatment with a mixture of copper(II) chloride and N,N′-dicyclohexylcarbodiimide (DCC).3 To examine this conversion we pursued a short synthesis of tuberculosinol from aldehyde 18. Thus methylenation of the aldehyde provided diene 19 in good yield. Subsequent regioselective hydroboration with 9-BBN and a palladium-mediated cross-coupling with (E)-3-iodobut-2-en-1-ol21 (21) (Scheme 4) provided tuberculosinol (16). On exposure to catalytic copper(II) chloride tuberculosinol (16) was converted into nosyberkol (15) (Scheme 4). We found that addition of DCC is unnecessary for this transformation and propose that this is a Lewis acid-mediated allylic transposition.22 In addition, we observed elimination of the allylic alcohol to provide a mixture of dehydrated products (34% as an E/Z mixture at the C12–C13 olefin); however, formation of the proposed structure 1 was never observed.
In conclusion, a rapid and stereodivergent synthesis of the core of the originally proposed structure of edaxadiene (1) raised questions regarding its assignment. Re-evaluation of the spectral data led to the proposal that edaxadiene is actually nosyberkol (15), a known diterpene previously isolated from Raspailia sp. Furthermore, an independent synthesis of nosyberkol has unambiguously established this structural revision.23 This synthesis of nosyberkol should provide sufficient quantities of this compound to elucidate its role in the infectivity and virulence of M. tuberculosis.
Footnote |
† Electronic supplementary information (ESI) available: Experimental procedures and spectroscopic data; interview with Erik Sorensen. See DOI: 10.1039/c0sc00284d |
This journal is © The Royal Society of Chemistry 2010 |