B.R.S.M. When all you have is a hammer everything looks like a nail


Cyanthiwigins B, F, and G

Total Syntheses of Cyanthiwigins B, F, and G

Stoltz et al., Chem. Eur. J., 2011, Early View; [PDF] [SI] [GROUP]

DOI: 10.1002/chem.201100425

I briefly contemplated not covering this, as I covered the Stoltz group conquest of liphagal last month,[1] and this uses the same palladium-catalysed enantioselective decarboxylative alkylation as the key asymmetry creating step. However, this is such a clever and powerful demonstration of the methodology that I couldn't help myself.[2] This time the targets are the cyanthiwigins, a family of marine diterpenoids with over 30 members, all sharing a highly conserved 5-6-7 tricyclic core. The few members isolated in sufficient quantity for testing, and the cyathanes in general, have demonstrated a range of biological activities so new routes to these compounds could be useful.

The starting material for the synthesis was diallyl succinate which, upon treatment with sodium hydride, underwent a Claisen condensation followed by a Dieckmann cyclisation. Quenching with iodomethane and warming slightly gave the expected 1:1 mixture of diastereomeric alkylation products (one racemic, and one meso). Interestingly, these were readily separable chromatographically, and found to have very different physical properties - one being a clear oil, the other a fluffy white solid. When the 1:1 mixture obtained from the previous step was added to a premixed solution of [Pd(dmdba)2] (dmdba = 3,5-dimethoxydibenzylideneacetone) and (S)-t-BuPHOX then a separable mixture of the desired (R,R) product and the meso biproduct was obtained in good yield after simply stirring at room temperature.

This is quite an amazing reaction as it takes a mixture of three stereoisomeric starting materials and converts them to predominately the desired compound, essentially as a single enantiomer, and sets both of the stereogenic quaternary centres found in the cyanthiwigins. In order for this challenging reaction to work the catalyst was required to decarboxylate (to the enolate) each starting material at around the same rate (to prevent excessive kinetic resolution at this stage) and then to carry out the desired asymmetric alkylation, regardless of the configuration of the other stereogenic centre. There's a very detailed discussion in the paper of the potential problems and outcomes, which is well worth reading, but needless to say that catalytic asymmetric reactions that set two stereogenic quaternary centres on opposite sides of the molecule in the same step are uncommon for a reason. Given the awesome e.e. (99%) you might wonder why the d.r. is a bit poor (4.6:1), but instead of the e.e. being excellent in spite of the poor d.r., it's excellent because of the poor d.r. This is because if a single alkylation occurred with the incorrect stereochemistry then the e.e. of the product wouldn't be affected, as this 'one-right, one-wrong' product is meso (and therefore lowers the d.r. and not the e.e.). Only if two alkylations go wrong is a molecule of the incorrect enantiomer produced (and the e.e. lowered). As the catalyst control is good, this doesn't happen very much and the formation of the meso product acts as a buffer, sacrificing d.r. to keep the e.e. high.[3]

Unfortunately the diketone product proved unresponsive to attempts to perform Grignard addition, presumably as attack was impeded by the a-quaternary centres. It was eventually found that this compound could be converted to the vinyl triflate and that a Negishi reaction could then be used to couple this with a homoallyl iodide. RCM was effected using Grubbs-Hoveyda catalyst, then the vinyl boronate was added to perform a separate cross metathesis reaction on the other allyl chain in the same pot. Oxidative workup with sodium perborate gave the aldehyde that was then cyclised via the corresponding acyl radical onto the nearby alkene, using AIBN and tert-butyl thiol to give the 5-6-7 system. The required trans stereochemistry at the 6-7 junction was obtained, as hydrogen atom abstraction from tert-butyl thiol occurs under thermodynamic control. The new ketone was then be converted to cyanthiwigin F by formation of the corresponding vinyl triflate, followed by Kumada-type coupling with the higher order cuprate preformed from 2 equivalents i-PrMgCl and 1 equivalent CuCN.

The same ketone was also oxidised to the enone using a method I hadn't seen before developed by Jiro Tsuji (of Tsuji-Trost fame) back in the early 80's.[4] They don't say why they chose this over selenoxide/sulfoxide elimination, Ito-Saegusa, or Nicolaou oxidation. Addition of the organocerium reagent derived from iso-propyllithium to the enone carbonyl, followed by Dauben-type oxidative transposition of the tertiary allylic alcohol to the enone gave cyanthiwigin B. Finally, reduction of both ketones in this natural product, followed by selective reoxidation of the new allylic alcohol, and dehydration of the non-allylic alcohol gave cyanthiwigin G. Nice work!

1. It was also one of my favourite syntheses of May/June. Check it out.

2. Another possible reason to not do this is that essentially the same enantioselective synthesis of cyanthiwigin F was published by Stoltz in Nature back in 2008 (Nature, 2008, 453, 1228-1231). That was only a letter, though, and this full paper contains a lot of interesting discussion and extra work, as well as two extra natural products. And I didn't have a blog back then.

3. If you didn't follow that, check out the paper, as Stoltz's explanation is much better than mine, and has pictures and references. This is a bit like the Horeau principle for enantioenrichment by dimerisation, the best explanation of which that I'm aware of is somewhere in Warren and Wyatt's Organic Synthesis: Strategy and Control (the sequel to the Disconnection Approach). I think. If you don't have a copy of this to hand (I don't, as I'm still on holiday) then you can check out this free handout from an old Stoltz group literature meeting. The concept of dimerisation as an enantioenrichment technique is old, and works in a very similar way.

4. Tsuji et al., J. Am. Chem. Soc., 1982, 104, 5844–5846.There's a full catalytic cycle so if you're gonna look, try and write one down first, then check it!


Comments (6) Trackbacks (0)
  1. Wouldn’t a temporary kinetic resolution be irrelevant as long as the reaction eventually went to completion?
    Shouldn’t a yield of 64% & 14% be a d.r. of 4.6:1 (the Nature paper seems to have the same inconsistency)?
    The final step for F used Pd(dppf)Cl2 (at least in the Nature paper).
    Was CH2Cl really used to make G?
    Why would deuterated chloroform be used in the final step for G?

    • The CEJ paper has the name numbers and says it’s 4.4:1 as well. Didn’t think to check that. The other two things are just typing errors, thanks. Having thought about the KR a bit more I’d be inclined to agree that as long as the reaction goes to completion, and Stoltz never says it doesn’t, then it shouldn’t matter. When I had another look at the paper, though, I saw this sentence, “If a large disparity in the rate of decarboxylation existed between different stereoisomers of starting bis(beta-ketoester) 36, undesired kinetic resolution would influence the downstream stereoselective bond formation.” I don’t follow that, any idea what they mean?
      With regards to the CDCl3, I did wonder myself; I don’t think it’s a mistake as they actually say ‘deuterated chloroform’ in the paper, and it’s in the SI. They report that last step in the SI on 1.1 mg SM (4 micromoles) in 250 microlitres of solvent, which I recon probably isn’t enough to do the reaction in an NMR tube and monitor it that way. Anyone have any ideas?

  2. Incidentally, looking over the SI (free, but >30mb) I see that they’ve included actual copies of IR spectra for all new compounds, in addition to the NMR spectra. Who does that? It can’t actually be required by the Journal, can it? Last time I saw copies of IRs in a paper was probably Woodward’s synthesis of Reserpine, although I don’t read SIs as much as I should.

  3. I believe that CDCl3 is used in the final elimination reaction because, unlike normal chloroform from an ACS bottle, it doesn’t contain a stabilizer? Martin’s sulfurane doesn’t play well with stabilizers found in most chloroform.

    Grabbing a bottle of CDCl3 is probably a quick and easy way to get dry, unstabilized solvent for the reaction. Stoltz did the same thing in his synthesis of MLR-52.

    • That makes sense – all of the ACS grade chloroform in the lab where I work is stabilised with ethanol, which I guess isn’t likely to get on well with the sulfurane. And as you say, most of us dry (and deacidify) our CDCl3. Thanks for the insight!

      • Apart from ethanol, the other stabilizers that are commonly employed for chloroform are olefins, i.e. amylene. They should not interfere with elimination using sulfurane. I think they run the reaction in CDCl3 and with sulfurane (as opposed to using, for example, Mitsunobu without a nucleophile for the elimination) that its progress is easy to follow by NMR – the reagent and its sideproducts have only aromatic protons and a nice fluorine signal. The entire react mix could have been transferred from a flask into NMR tube (without evaporation) with some additional CDCl3.

Leave a comment

No trackbacks yet.