Total Synthesis and Stereochemical Reassignment of (±)-Indoxamycin B
There haven’t been many total syntheses recently that I’ve really wanted to write about in the last month or two, but I quite enjoyed both the Carreira offerings that appeared in Angewandte last Friday. After realising that I didn’t have time to write about both, I decided on this one as it reminded me of some chemistry I’d done myself a while back. I also enjoy it when total synthesis ends in reassignment, as it’s probably one of the more worthwhile outcomes of a synthetic campaign, and it makes a nice story.
The route began with a Birch alkylation of methyl 3,5-dimethylbenzoate using what amounts to protected iodomethanol. If you plan to have an ester survive Birch-type conditions, bigger is definitely better, and pivaloyl is a smart choice here. LiAlH4 was then used to cleave the pivaloate and simultaneously reduce the starting methyl ester to give the dienediol. The diol was then protected as the acetonide and then an allylic oxidation was performed using t-butylhydroperoxide and catalytic palladium-on-carbon. The group then treated the rather unstable dienone with the allyltitanocene complex preformed from the allyl ether, titanocene dichloride and butyllithium, which allowed selective 1,2-addition to the rather hindered ketone in reasonable yield. This was followed by an anionic oxy-Cope reaction, where the enolate formed was trapped as the TES silyl enol ether for use in the next step where a rather neat palladium mediated oxidative cyclisation was used to form the second ring of the natural product.
Next, the group cleaved the acetonide with a bit of acid and epoxidised the cyclopentene double bond which under simple substrate control occurred from the top, convex face. They didn’t isolate the epoxide, however, as it was immediately opened by the nearby alcohol to give the diol shown. The primary alcohol was then selectively protected with a TBS group, and the secondary oxidised to the ketone using the Dess-Martin reagent. O-alkylation of the ketone with propargyl bromide, KH and some -cr-6 then gave the propargyl Claisen precursor that was rearranged by heating with a catalytic amount of an odd looking gold(i) oxo complex to keep the temperature required low. A fairly impressive chemo- and stereoselective reduction of the cyclohexenone carbonyl was then achieved in good yield using Super Hydride, and the resulting alcohol was cyclised onto the allene using a bit more gold to form the final ring in the natural product with fair diastereoselectivity.
The group then went on to take this intermediate through to the original structure of the natural product, and then found that the data didn’t match, causing them to propose a revised structure and have another go.
This second approach began with a Mukaiyama hydration of the terminal double bond, to give a 1:1 mixture of diastereomeric alcohols that were oxidised to the methyl ketone. I’d never seen this sequence used to replace a Wacker-type oxidation, but it's a sensible tactic here, as adding palladium to allyl ethers can have undesirable consequences. This new ketone then underwent Wittig olefination with excellent chemoselectivity but highly variable stereoselectivity. Fortunately, the reaction did at least consistently favour the desired E- product, and the two isomers were separable chromatographically. The α-ketoether was then cleaved with samarium diiodide, and the alcohol released was oxidised to the corresponding aldehyde, both steps occurring in essentially quantitative yield. An HWE reaction was then used to install the unsaturated ester sidechain and the cyclopentanone was converted to a cyclopentene by reduction of the ketone and dehydration of the resulting alcohol under Burgess conditions. Finally, hydrolysis of the methyl ester gave the natural product in excellent yield. This time, all data was found to match and group could finally relax!
Addenda and Tangential Information
1. I actually have loads of ideas for posts, but I’ve just entered the last month of my PhD funding, I’m still in the lab, and things are a bit hectic. Also, the graphical abstract for the other paper contains the word 'linchpin'.*
2. I’ve always found that for Birch alkylations (not so much reductions, which are easier), that I’ve had to condense ammonia from the cylinder into an empty flask, dry it with sodium and then distil it again, into my actual reaction flask. Maybe our ammonia is just a bit damp, but using it out of the cylinder tends to result in an inseparable mixture of alkylated and reduced products.
Unfortunately, ammonia takes a really long time to evaporate, and if one simply removes the cooling bath from the drying flask and waits then it can take a couple of hours for just 100mL to distil over into the reaction vessel. The last time I had to do one of these was a day when the group had planned to go out for lunch, and fifteen minutes before everyone left I found myself furiously distilling ammonia from the drying flask into my reaction vessel with a heat gun so I’d get the reaction on before I left. Not recommended.
3. I used to always do these post-Birch allylic oxidations with chromium based reagents but I’ve recently found this Corey allylic-oxidation (Org. Lett, 2004, 4, 2727) to be much simpler, higher yielding and more scalable. Also, it produces significantly less chromium waste.
4. Although I’ve done a fair few of these myself (on aryl systems) I didn’t know until reading this paper that the transformation actually has name: the Saucy-Marbet reaction. Which reminds me of a girl from my undergrad. In this case the reaction couldn’t be done without a catalyst as the high temperatures required degraded the substrate. I’ve had to heat these at 200+ ºC in 1,2-dichlorobenzene before, and can attest they often require quite a kicking to go thermally.
*I'm joking here