Construction of the polycyclic
skeletal architecture of the mexicanolide class of limonoid (tetranortriterpenoid)
natural products by chemical means is fraught with difficulty, even by today’s
synthetic standards. In 1971, Connolly and co-workers converted an alternate
tetranortriterpenoid (7-oxo-7-deacetoxykhivorin) into mexicanolide in the
course of a 7-step partial synthesis. Over the next 40 years, little or no additional tactical refinements or conceptual
advancements to facilitate access to the mexicanolide system were reported in the
synthetic literature. Then, in 2012, Williams’ laboratory at the University of
Queensland reported the first total synthesis of (-)-mexicanolide,
as well as the related bicyclononanolide, (-)-khayasin.
Similar to Connolly’s pioneering work, Williams’ synthetic strategy exploits
putative biogenetic relationships existing between several known limonoid
natural products, including azedaralide (3), cipadonoid B (9) and proceranolide
(12).
Williams’ biomimetic synthetic
approach to the mexicanolides dictated condensation of the advanced optically
active building blocks, azedaralide (3) and the methyl cyclohexenol ether 7. To
begin, the asymmetric synthesis of 3 was enabled by implementation of a chiral
borane (DIP-Cl)-controlled aldol condensation, which established the vicinal
stereocenters resident in the beta-hydroxy enone 2. Intermediate 2 was then elaborated
into the natural product 3 by a three-step sequence that was previously
executed by the same group during their synthesis of racemic azedaralide.
Another chiral borane-mediated
enantioselective aldol reaction produced the hydroxy ketone 5, which underwent
base-promoted cyclization to give 6 with minimal stereoerosion. The cyclic
enone 6 was then converted into the requisite methyl enol ether 7 using methyl triflate.
Next, a ketal-Claisen rearrangement
involving the advanced optically active building blocks 3 and 7 was utilized to
secure the all-carbon quaternary center resident in (-)-cipadonoid B (9). A computational investigation of the mechanistic
reaction pathway leading to 9 (3 + 7 à
9) guided the optimization of the stereoselectivity and efficiency of this
highly complex, one-pot, cascading transformation. The alpha-oxirane (10),
derived in one step from cipadonoid B, then underwent reductive single-electron
fragmentation to generate a reactive enolate species (11). The enolate 11 then
engaged the proximal exocyclic olefin in an intramolecular 1,6-conjugate addition
that accomplished a 6-endo-trig
cyclization to fashion the mexicanolide ring skeleton of proceranolide (12).
Again, the daunting complexity of the overall transformation (i.e. 10 à 12) must be
acknowledged in the assessment of the seemingly moderate efficiency (30%
isolated yield) of the process. The advanced intermediate 12 was successfully
converted into the tetranortriterpenoid natural products (-)-mexicanolide and (-)-khayasin by straightforward redox adjustment and esterification,
respectively. It should be noted that the campaign by the Williams laboratory that
is described above constitutes one of only three total synthesis studies that
culminated in the completion of an intact limonoid natural product. In addition
to Williams’ limonoid research efforts, Steven Ley’s 79-step, 22 year total synthesis of azadirachtin and E. J. Corey’s total synthesis of azadiradione
round out the remainder of the (to the best of my knowledge) comprehensive
list. This astonishing fact stands as a testament to the formidable scientific
challenges associated with the chemical synthesis of bioactive limonoids.
