B.R.S.M. Yield isn't everything

Kibdelone C

Porco et al., J. Am. Chem. Soc., 2011, Article ASAP; [PDF] [SI] [GROUP]

DOI: 10.1021/ja203642n

Ready et al., J. Am. Chem. Soc., 2011, Article ASAP; [PDF] [SI] [GROUP]

DOI: 10.1021/ja204040k

Here’s an odd occurrence; two quite different syntheses of the natural product kibdelone C in appeared in the JACS ASAP on the same day last week; one by the Porco group and another by Ready and coworkers. Each acknowledges the other for sharing details of the work before publication, so I guess the authors were less surprised than I was. There are a number of kibdelones, all of which are quite similar and tend to interconvert on standing. They boast antibacterial, antinematodal and anticancer activity. The mode of action isn’t known, but they look likely to bind nucleic acids.

Porco

Retrosynthetically, both groups decided divide the molecule into three fragments by disconnecting through the C and E rings, but the chemistry employed in their unions is very different. I'll start with the Porco group, who worked alphabetically, first preparing the AB ring system, then forming the C ring by attaching styrene D. They published this chemistry earlier in the year, but it’s quite interesting so I’ll summarise it here anyway. Looking at the quinone and styrene fragments above, you could be forgiven for expecting a Diels-Alder approach here, and indeed this was exactly what the group intended. A (retrosynthetic) plan never survives contact with the enemy, however, and despite some literature precedent, they were unable to get any [4 + 2] product. Fortunately, after screening some fairly expensive looking Lewis acids (Au(III), Pt(IV) and In(III)) they found that the same fragments could be combined in an ionic arylation. PtBr₄ proved to be by far the best, performing best with 2 equivalents of water for each equivalent of catalyst, implicating the diaqua complex as the actual active species. The group cites some interesting literature, explaining that the acidity of water molecules bound to Pt(IV) centres is greatly increased, and that crystal structures of Pt aqua complexes show that the bound water molecules can hydrogen bond to nearby ethers (which leads to some interesting speculation about a mechanism for the reaction). In any case, the reaction performed well enough, giving a workable 55% yield on multigram scale. Photoelectrocyclisation and desilation with TBAF neatly delivered the ABCD fragment.

Next the group set about preparing the F ring fragment, whose attachment would also lead to the construction of the E ring. Containing all three of the molecule's stereogenic centres in a penta-substituted cyclohexene, you might expect some interesting chemistry here, and you’d be right. The group decided to start from the known homochiral aldehyde shown below, beginning with a chelation controlled Zinc acetylide addition to set the second stereogenic centre. Some standard transformations (oxidation level and protecting group adjustments) gave the aldehyde required for the cyclisation. Upon treatement with 2 equivalents of magnesium iodide the halo-Michael addition – intramolecular aldol cascade worked beautifully, giving the F ring cyclohexene in 78% on a multi-gram scale, with a small amount of a diastereomeric byproduct.

Additional chelation between the α-benzyloxy group and the aldehyde in the transition state above by a second magnesium atom may explain the preference for the desired diastereomer (I can’t draw this, and I’m not going to try – see the paper for further discussion). Unfortunately the F ring fragment above didn't undergo the planned addition-elimination, and some protecting group adjustments were needed before it could be united with the ABCD unit. Thus, when the adjusted iodoacrylate below was heated with the ABCD fragment and tribasic potassium phosphate in DMSO, the key addition-elimination occurred as expected, in a modest yield of 44%.

The final cyclisation reaction to forge the E ring pyranone proved challenging, as the use of standard Friedel-Crafts reagents (polyphosphoric acid, Eaton’s reagent etc.) was thwarted by the ‘disappointing’ acid lability of the ester. The group countered by preparing the free acid, which they planned to convert to the acid chloride, then cyclise with aluminium trichloride. However, it was found that when the acid was treated with cyanuric chloride and pyridine to form the acid chloride, the cyclised compound was isolated instead, even in the absence of a Lewis acid. The group theorises that reason for the ease of this transformation is a result of the formation of a ketene intermediate, through participation of the oxygen lone pair, which can undergo a rapid 6π electrocyclisation to complete the tetrahydroxanthone.

Note: X could be chloride, pyridine or even the activated ester formed by reaction with cyanuric chloride. Finally, the acetonide was cleaved and the B-ring was demethylated by oxidation to the quinone, followed by reduction with sodium dithionite to the requisite hydroquinone. Although a little redox inefficient, this method is far milder than Lewis acid mediated or nucleophilic demethylation, and was also selective for the B ring over the D ring, when conducted under acidic conditions. The group suggests that in the presence of acid (they used 10 equivalents of AcOH)  the E ring exists in its pyrylium form, by protonation of the xanthone carbonyl, strongly disfavouring oxidation of the neighbouring D ring.

Ready

In contrast to the Porco ABCD + F endgame, the Ready group prepared the AB and DEF fragments and then linked them to form the C ring last. The AB ring was easily prepared from the amide shown, which was derived from the known carboxylic acid. The alkyne linchpin which would be used to provide two of the C ring carbon atoms was introduced by Sonogashira reaction with TIPS-acetylene, followed by desilylation with TBAF.

My favourite part of the Ready synthesis is the way the group were able to exploit the pseudo-C₂ symmetry of the F ring to disconnect the vinyliodide back to the bis(silyl enol ether) below. Now that's non-obvious. The cornerstone of this sequence was Andy Myers' methodology for the transformation of silyl enol ethers into monoprotected trans-diols.

Thus, the bis(epoxide) formed from the bis(silyl enol ether) using Shi's catalyst was opened reductively with borane THF complex to give the C₂ symmetric diol shown. This epoxide opening step is believed to proceed via a carbocation as shown above, and even more interestingly, still works really well even for acyclic epoxides. The rotation around the central sigma bond is slow relative to the hydride transfer so decent diastereoselectivity is still obtained (see the Myers paper for more detail). As the absolute configuration of natural kibdelone C t wasn't known until Porco made it Ready had to pick a Shi catalyst, but unfortunately he guessed wrong and ultimately produced the antipode of the natural product. However, it would be possible to repeat the synthesis with the other Shi catalyst and make the correct enantiomer of the target just by varying this step. Monotosylation of the diol and Swern oxidation gave a ketone, which, when treated with iodine and pyridine, suffered E1cB loss of the tosyloxy group to give the α,β-unsaturated ketone that then underwent Baylis-Hillman type iodination. Luche reduction under substrate control then gave the complete F ring fragment.

After deprotonation of the alcohol with methyllithium the vinyl iodide underwent halogen-lithium exchange with t-BuLi, and was added into the D ring aryl aldehyde (prepared by MOM-protection and formylation of a known compound). Oxidation to the enedione using the Dess-Martin periodine, followed by treatment with perchloric acid causing TBS and MOM deprotection, conjugate addition and acetonide formation (quite possibly not in that order) gave the DEF tetrahydroxanthone. This was Sonogashira'd onto the AB acetylene (after reprotection of the secondary alcohol as a MOM ether) in good yield and the triple bond was reduced to the alkane. The B ring phenol underwent ortho-iodination, and was subsequently protected with a BOC group to provide the substrate for the crucial arylation reaction expected to close the C ring and complete the hexacyclic ring system of the natural product. Although an impressive 63% yield of the completed kibdelone skeleton was eventually obtained for the C-H - halide coupling, the reaction apparently required extensive optimisation, with deiodination, loss of protecting groups, and aromatisation of the F ring all competing with the desired outcome. From here, chlorination and 3 deprotection steps gave the unnatural enantiomer of the natural product.

So, two very different syntheses of the same natural product, with plenty of cutting edge work alongside the usual handful of named reactions in each. In both cases some very interesting chemistry was used to set the F ring stereocentres, and the care taken in the last steps with these difficult to handle natural products was impressive. Nice work!