Update 26/11/11: Link to newly available SI now added!
Update 30/10/11: Also, Tot. Syn. now has a post on this.
Update 29/10/11: Covered in more colours over at synthetic nature. Check it out!
Carreira et al., Angew. Chem., 2011, Early View
Here's an interesting synthesis from a couple of weeks ago out of the Carreira lab - the first synthesis not only of daphmanidin E itself, but of any alkaloid of its family and skeleton. The daphmanidins constitute a recent addendum to the already fairly diverse class of alkaloids isolated from Daphniphyllaceae, with today's target only reported 5 years ago. Presumably, exciting features like "moderate vasorelaxant activity on rat aorta" and that well nested central bicyclo[2.2.2]octane ring make them tempting targets...
The racemic starting material is obtained from diethyl succinate, with one and two pot versions reported. There are also several options for its resolution including chemical and chemoenzymatic methods. According to a footnote "the resolution via diastereomeric hydrazones is detailed in the Supporting Information", but I can't seem to find any SI accompanying this paper! If I'm missing something, please point it out in the comments!
Monoprotection of the diketone proved difficult, so it was first doubly protected using propane-1,3-diol and p-TSA, and one of the equivalent dioxanes was cleaved by acetal exchange with acetone and more p-TSA to give the monoacetal in good yield. This compound was then converted to the vinyl triflate and coupled with the alkyl borane obtained from hydroboration of the TBDPS protected allyl alcohol shown. Next came a rather surprising diastereoselective hydroboration, with the borane approaching from the convex face, syn to the bicyclo[2.2.2]octane bridge. Interestingly, during early studies the group found that the starting diketone could be converted selectively to the mono vinyl triflate and coupled in good yield without the need for the dioxolane protection just discussed. However, unless the remaining ketone was protected for the hydroboration step then the undesired diastereomer was obtained. Even when the borane could be directed to the correct face, the reaction was still sluggish, and had to be run with excess borane at high concentrations, leading to partial reduction of one of the esters. This proved inconsequential, however, as the group had intended to reduce both esters in the next step anyway using DIBAL and this reaction was simply performed on the crude product from the hydroboration. Finally, an acetal adjustment and conversion of the primary alcohol to the benzoate gave the funtionalised bicyclo[2.2.2]octanone shown.
When it came to functionalising the remaining ketone α-position the group employed a slightly unusual double Claisen rearrangement strategy. The reason for this was that initial studies had found that the hindered nature of the relevant methylene group caused O-alkylation to be strongly favoured (and C-alkylation to be very difficult), requiring a less direct approach. Thus, treatment of the ketone with KHMDS, 18-crown-6 and the allyl tosylate shown (whose preparation is described in the SI, wherever that is) smoothly effected O-alkyation of the enolate generated. The first Claisen rearrangment proceeded in good yield and diastereoselectivity upon heating in nonane. O-alkylation with allyl bromide proceeded easily as before, but due to the hindered nature of the bond being formed upon rearrangement the yield for the second Claisen was a lot lower. Selective hydroboration of the less hindered hindered olefin, followed by oxidative work up and protection gave the acetate. The sidechain TBDPS group was removed using TBAF and the resulting primary alcohol was dehydrated using the popular Grieco protocol. Next, the acetonide was cleaved and it became necessary to differentiate between the primary and secondary alcohols present. This was done by temporarily protecting the primary alcohol as the TMS ether, MOM protecting the secondary alcohol, and then removing the TMS, allowing the primary alcohol to be oxidised to the aldehyde. I wonder if preferential oxidation of the primary alcohol using something like the very selective Piancatelli conditions would have worked instead, but I guess the 90% yield obtained over the whole sequence would have been hard to beat, even if it's a little inelegant. Condensation of the new aldehyde with nitromethane gave the expected nitro olefin and set the stage for the potentially difficult asymmetric introduction of the methyl group that would ultimately be incorporated into natural product's dihydropyrrole ring.
Initial studies on the conjugate addition of various methyl carbanion nucleophiles found that the natural preference of the molecule (under substrate control) gave various mixtures of epimers as 5:1 - 9:1 mixtures in favour of the undesired compound. Fortunately, the group found that this preference could be overwhelmed using the catalyst formed from [Cu(OTf)]2·toluene and the known Hoveyda phosphine ligand with dimethylzinc as the nucleophile. This reaction proceeded in a spectacularly high yield of 90%, giving a 5:1 mixture of epimers in favour of the desired compound, and illustrating the power of (catalytic) reagent control for when your molecule just wants to do the wrong thing. Reduction and Boc protection of the nitro group followed by a double ozonolysis gave the ketoaldehyde, and the aldehyde was then chemoselectively reduced using sodium triacetoxyborohydride in acetic acid-THF. The resulting primary alcohol was then converted to the mesylate and thence to the iodide under Finkelstein conditions. Finally, the MOM-protected β-hydroxyl group was eliminated to give the cyclopentenone needed to investigate construction of the cycloheptene ring. After the failure of various Pd, Cr and Sm methods the group turned to Cobalt, specifically the cobaloxime shown below. At first, the iodide was treated with 1.1 equivalents of this under a sunlamp, giving the Heck-type product in an amazing yield of 95%. Somehow still wanting more, the group then developed an improved version requiring only catalytic amounts of the cobaloxime and 1.5 equivalents of Hünig's base, which proceeded in essentially the same yield.
From here the remaining two rings were formed without incident. The acetate protecting group was removed and the unmasked alcohol oxidised and cyclised onto the nearby ketone in an intramolecular aldol condensation. The aldehyde was then converted directly to the methyl ester using the somewhat uncommon Corey-Ganem-Gilman oxidation in methanol. Swapping the benzoate for an acetate, followed by treatment with TFA (to remove the Boc group) and heating in ethanol formed the dihydropyrrole ring. Finally, MOM deprotection with Ph2BBr gave the natural product in good yield. A great piece of work from Carreira and coworkers, with many clever solutions to problems met along the way!
1. This is the silliest biological activity I've seen reported for a while, after panacene's (Tot. Syn.) "shark antifeedent" qualities. After testing against cancer cell lines and other biological targets and coming up negative I guess isolation chemists keep going until they can report some biological activitity. "No cytotoxicity? Hmm. Let's see what it does to this rat aorta, obese whale shark and bipolar walrus. We can't just leave it blank..."
2. Although still not common, B-alkyl Suzuki coupling (especially sp3-sp2) are quite possible. For an excellent free treatment of the subject see the Macmillan group meeting presentation. For a slightly older and considerably less free treatment, the Danishefsky/Trauner review in Angew. Chem., 2001, 40, 4544 is also good and explains the problems which need to be overcome. In this case, triphenylarsine (a weak ligand very common in tricky Stille reactions) was needed to avoid the considerable reductive detriflation caused by more common phosphine ligands.
3. An excercise for the reader: try and draw the transition state leading to the desired product and then check it against Carreira's in footnote 13.
4. On reflection it seems kind of odd to me that the sole function of this group was to direct the Claisen rearrangement. OMOM is not your typical directing group (why not something bigger?), and it's a shame the group took 4 steps from (R)-cyclohex-2-enol to get make the allyl tosylate with this centre set only to bin it later. Ah well, if it works...
5. Although they call this a Co-Heck, and it's equivalent to that transformation, I don't think this is a good name; check out the original paper (J. Org. Chem., 1995, 60, 6635) for tasty mechanistic discussions. Basically, it's all about photolytic homolysis of the Co-Sn bond and ensuing radical fun.
6. I have a half written article about this somewhere which may appear later in the week depending on how much time I have.