Routes to Advanced Intermediates in the Synthesis of Tetracarbocyclic Sesquiterpenoids Daphnenoid A and Artatrovirenols A and B

A short route from dihydrocarvone is described, which led to the tetracarbocyclic core common to artatrovirenol A and B and daphnenoid A. A variant of this route afforded guaia-4,6-dien-3-one (from Enterospermum madagascarensis) and its epimer. From 2-(2-oxoethyl)furan, a 15-step sequence then delivered the complete carbon skeleton and all functionality necessary for daphnenoid A. Key steps in the route include diastereoselective intramolecular oxidopyrylium cycloaddition, oxa-bridge cleavage under “push–pull” conditions, and intramolecular Diels–Alder cycloaddition.

T he sesquiterpenes and their oxidized derivatives, the sesquiterpenoids, form a major class of natural products with structures based on 1.5 (sesqui-) C 10 -monoterpene units, the first established being those of α-santalol 1 and farnesol. 2heir wide-ranging biological properties and vast range of structural types have attracted the efforts of synthetic chemists for more than a century, with many of these efforts patterned on biosynthetic speculations. 3Recently, a new caged-ring sesquiterpenoid subclass has emerged in two papers reporting the structures (Figure 1) of artatrovirenols A 1 and B 2 from Artemisia atrovirens, 4 and daphnenoid A 3 from Daphne penicillata. 5The same tetracyclic ring system is also found in the prenylated phloroglucinol garcinielliptone enantiomers HG and HH from Garcinia subelliptica. 6etracarbocyclic frameworks are rare in sesquiterpenoids, those containing cyclopropanes excepted; therefore, this new sesquiterpene subclass is of interest not least from a biosynthetic standpoint.Both Chen's 4 and Song's 5 reports associate these novel structures with the guaiane system 4.The new framework 5 can be derived from 4 both conceptually and biosynthetically by connecting the isopropenyl Δ 11,12 carbons to C-1 and C-4, respectively.Building on our recent studies of simple guaianes, 7 our group set out to explore the synthesis of daphnenoid A based on an intramolecular Diels−Alder (IMDA) cycloaddition or an equivalent process from an appropriate guaianoid precursor.Very recently, Zhu's group reported the total synthesis of (−)-artatrovirenol A by an unrelated synthetic strategy, 8 which has prompted us to disclose our current progress in this area.
Retrosynthetically, it was envisaged that the new ring system would be constructed from enone 7 (Scheme 1) by using a silylation-induced (Mukaiyama) Michael/Michael-type reaction via dienol ether 6.In this scheme, the appropriately configured protected 3°-alcohol in this enone would arise by elimination of the 1,7-oxa bridge in 8, the product of intramolecular oxidopyrylium cycloaddition via 9. Completing the analysis led back to a 2,5-disubstituted furan of general form 10, although it was appreciated that the combination of a furfurylic alcohol, a β,γ-unsaturated ketone, and a methylene flanked by both electron releasing (furan) and electron withdrawing (carbonyl) functionality would render this specific precursor somewhat fragile.
Before embarking on the route outlined above, a preliminary investigation of the viability of the key formal cycloaddition step (cf.7 → 3) was undertaken (Scheme 2).The known ring-   13 the crude reaction product consisted mainly of the desired tetracyclic ketone 15.This ketone was found to be unstable toward purification by chromatography on silica gel; therefore, ketone reduction was carried out, which enabled a pure sample of alcohol 16 to be obtained whose structure was confirmed by single-crystal X-ray diffraction. 14dapting this route to incorporate the C-15 methyl group required for the complete sesquiterpene skeleton was not achieved.Although a suitable substrate 19 was prepared from the same keto-ester 11, attempts to access the [4 + 2]cycloadduct failed, and the only tractable result was alkene isomerization to deliver known 15 dienones 20 (a sesquiterpenoid from Enterospermum madagascarensis) 16 and 21.
It was concluded that the dienophile in the C-15 methylbearing enone would require electronic activation, and so, the route outlined in Scheme 1 was initiated.The first phase of the synthesis proceeded smoothly, with little in the five steps 17−19 to hydroxypyrone 26 (Scheme 3) requiring comment.The anti-crotylation 20 product (S2, see Supporting Information) derived from aldehyde 22 was targeted in order to allow the two substituents in the projected cycloadduct to be both cis and exo-disposed, 21 thus easing the course of the cycloaddition.The use of classical methods for initiating formation and cycloaddition of the oxidopyrylium derived from 26 were unproductive or inefficient; however, Suga's mild conditions using Boc-anhydride to activate the hydroxyl group, in combination with triethylamine as catalyst, 22 produced the separable diastereomeric exo-cycloadducts 27 and 28 in ∼80% combined yield.In this reaction, the major isomer is that expected from Sammes' studies of related reactions. 23onjugate addition of lithium dimethylcuprate to the more exposed α-face of the enone, followed by desilylation and oxidation, afforded tricyclic diketone 30, from which it was envisaged that mild treatment with basic or Lewis acidic reagents would induce enone formation with cleavage of the C-1−O bond (→ 31).In the event, no reagent combinations were able to achieve this elimination, presumably because any ring-opening remained hidden by rapid reclosure of the soformed 3°-alcohol.
Taking inspiration from Mascarenãs' "push−pull" approach to oxa-bridge cleavage in related structures, 24 the steps from cycloadduct 27 were reordered to enable the two ketones (in 32) to be distinguished.Thus, the "pushing" silyl enol ether functionality was introduced prior to methyl 1,4-addition, and then, the Lewis acidic "pulling" conditions were applied to soformed 33 with the inclusion of acetic anhydride to capture liberated 3°-alcohol and prevent its recyclization.This was effective, and the acetoxy enedione 34 was accessed in readiness for the proposed IMDA-equivalent step.Again, forward progress was thwarted since all attempts to effect cycloaddition using variants of the conditions used for the formation of tetracycle 15 led to decomposition.Here, reaction mixtures usually developed a purple coloration that was attributed to azulene formation by acetate elimination and aerial oxidation.
To obviate the need to add activating reagents for the IMDA cycloaddition and reduce the likelihood of azulene formation, a preformed diene component was targeted and the oxidation state of the substrate was lowered (Scheme 4).First, the C-8 ketone was reduced, and the resulting alcohol acetylated during the β-elimination step to form the enone (→ 35).Luche reduction of the C-3 ketone then gave cyclopentenol 36.For the dehydration step, Burgess reagent was chosen based on its preference for a syn-elimination mechanism; 25 from this, a major cyclopentadiene regioisomer 37 was obtained.Once more, the incorrect placement of the diene was not envisaged to be problematic and, indeed, heating this substrate in toluene (in a pressure tube) initiated clean IMDA cycloaddition, with the tetracyclic cycloadduct 38 this time being perfectly stable toward chromatographic purification.
From this point, a potential end-sequence to daphnenoid A was projected: (i) double deacetylation, (ii) oxidation of the C-8 alcohol to the required enone, and (iii) formation of the C-3 carbonyl group by either hydroboration/oxidation or Wacker oxidation.In practice, methanolysis of triester 38 gave pentacyclic lactone 39, similar in structure to artatrovirenol A. Heating this lactone with triethylamine in methanol established an equilibrium with diol 40, allowing a sample of the latter to be separated.An attempt to access the C-8−C-10 enone directly with IBX 26 led to decomposition, but under milder conditions, ketone 41 was formed, which is currently our most advanced intermediate en route to daphnenoid A. 27,28 The synthetic chemistry summarized in Schemes 3 and 4 comprises a 15-step sequence to tetracyclic intermediate 38, from which some redox adjustments remain necessary to reach daphnenoid A. It is, however, conceivable that a variant of the route to form compound 15 (Scheme 2) could deliver the complete carbon skeleton more quickly, from which chemical and enzymatic C−H hydroxylation 29 would then offer access to this interesting class of tetracarbocyclic sesquiterpenoids and their analogues.Efforts to that end are ongoing in our laboratories.

Scheme 4 .
Scheme 4. Completion of the Tetracyclic Core and Progress towards Daphnenoid A Organic Letters