Perhaps not unexpectedly, given last year's exciting breakthrough syntheses, more synthetic work on the weltwitindolinones appeared in the literature last week, so I think an update is in order. As a reminder, the first synthesis of a bridged bicyclic member of this family came in March last year when Rawal reported a neat racemic route to N-methylwelwitindolinone D isonitrile, emerging as the winner of a 15 year race between plenty of well known synthetic chemists. This was rapidly followed by Garg, who published an excellent synthesis of (-)-N-methylwelwitindolinone C isothiocyanate in August, which I covered here.
Well, two more back to back JACS papers, one each from Rawal and Garg have just appeared, and I'll summarise both here. Garg's describes improvements to the key step in his previously reported route to (-)-N-methylwelwitindolinone C isothiocyanate, along with the synthesis of (-)-N-methylwelwitindolinone C isonitrile and a couple of the so-called 'oxidised welwitindolinones'. Rawal's contains an asymmetric version of his previous racemic route, allowing access to (-)-N-methylwelwitindolinone C isothiocyanate/isonitrile, along with one of the 'oxidised welwitindolinones'. As Rawal’s has the most new chemistry of the two we'll look at that first.
1. A Unified Route to the Welwitindolinone Alkaloids: Total Syntheses of (-)-N-Methylwelwitindolinone C Isothiocyanate, (-)-N-Methylwelwitindolinone C Isonitrile, and (-)-3-Hydroxy-N-methylwelwitindolinone C Isothiocyanate
I didn't actually cover Rawal's first paper (DOI: 10.1021/ja201834u) on these compounds, not because there was anything wrong with it, but because I didn't have a blog at the time it came out. It turns out that this new route uses the same first 7 or so steps, although beginning from an enantioenriched starting material where the previous paper used a racemic one. For completeness I'll include the shared steps here so you can see the whole thing from start to finish.
The route began with cuprate addition to the protected γ-hydroxyketone shown, followed by quenching with trifluoroethyl formate to form the β-ketoaldehyde that was then methylated using dimethyl sulfate. The starting material was available in an enantioenriched form by lipase resolution of the corresponding acetate, which proceeded in high ee. The ketone was then converted to the silyl enol ether under standard conditions and coupled with the N-methylindole alcohol shown in the presence of TMSOTf to give a single diastereomer of the product. Initially, the group used an N-arenesulfonyl protected indole derivative, but this gave much lower yields in the coupling, presumably because the indole nitrogen was much less able to stabilise the benzylic-type cation. Being able to use the N-methyl indole directly also shaved a couple of steps off the route, including a non-trivial deprotection and a methylation later on. Next came the crucial formation of the indole-bridgehead carbon-carbon bond, which was performed by a palladium mediated arylation of the potassium enolate of the β-ketoaldehyde. At the time, no examples of this reaction had yet been reported, making the 73% yield for formation of this challenging quaternary centre all the sweeter. Finally, removal of the TBS group with hydrofluoric acid and oxidation of the unmasked alcohol with Dess-Martin periodinane gave the diketone.
Now the group deviated from their previous route, as in order to access welwitindolindone C derivatives C-13 carbonyl needed to be converted to the corresponding vinyl chloride. First, the more electrophilic aldehyde was reduced, and the ketone could then be converted to the hydrazone by heating with hydrazine and acetic acid in ethanol. This hydrazone was then oxidised/halogenated by NCS in pyridine to give the vinyl chloride in good yield (especially considering the presence of the indole). Interestingly, this reaction proved inordinately sensitive to trace transition metal contaminants, and high yields could only be obtained with brand new glassware and stirrer bars. Next, the indole was converted to the oxindole using magnesium monoperoxyphthalate, a slightly milder peracid than the more commonly used m-CPBA. The primary alcohol was restored to the aldehyde using Dess-Martin periodinane and then converted to the oxime (as an inconsequential mix of diastereomers that was used directly in the next step). Next came an unusual transformation: the conversion of this mixture of oximes to the corresponding isothiocyanate, something I'd never seen before Rawal's previous welwitindolinone paper. Thus, treatment of the oximes with NCS, followed by triethylamine and propylenethiourea gave the first natural product in great yield.
This natural product could then be converted into two more, each taking one further step. Thus, desulfurisation with Mukaiyama's oxazaphospholidine gave N-methylwelwitindolinone C isonitrile or, alternatively, treatment with KHMDS followed by the Davis-type N-benzenesufonyl oxaziridine gave 3-hydroxy-N-methylwelwitindolinone C isothiocyanate. Nice work.
2. Total Synthesis of Oxidized Welwitindolinones and (-)-N-Methylwelwitindolinone C Isonitrile
The key steps in the endgame of Garg's previous synthesis are shown below. The group's clever solution to the problem of functionalising the tricky bridgehead C-11 position began with the diastereoselective reduction of the cyclohexenone carbonyl followed by its converted to the primary carbamate. This was then oxidatively converted to the acyl nitrene that underwent CH insertion at C-11. Unfortunately, as awesome as this step was, it was disapprovingly low yielding, giving just 33%.
A lot of people would be happy with such a yield, but as the Garg group had planned further studies in this series they needed more material and so had some optimising to do. Analysis of the reaction mixture found that the other major product of the reaction (at 25%) was simply the ketone at the start of the previous scheme. They reasoned that this was probably formed by unwanted CH insertion at C-10, followed by hydrolysis of the hemiaminal back to the ketone. As optimisation of reaction conditions failed to minimise this, the group decided to alter the starting material.
Thus, the troublesome hydrogen atom was replaced with something that would undergo CH insertion much more slowly, but could be installed and removed as easily as the hydrogen was in the route above, without introducing any extra steps. Any guesses? That's right... deuterium! Although, as Garg remarks in the paper "the strategic use of a deuterium kinetic isotope effect in total synthesis is rare" this is exactly what the group did, and it worked marvellously, almost doubling the yield to 60%. Introduction of the deuterium was done with Super Deuteride (unsurprisingly, the deuterium analogue of Super Hydride, i.e. LiEt3BD), actually offering a slightly improved yield over the DIBAL procedure originally reported. After the nitrene insersion, the same hydrolysis conditions as before were used to cleave the cyclic carbamate and the alcohol was then oxidised back to the ketone, removing the deuterium atom now that its work was done. Interestingly, the group used the Dess-Martin reagent here in place of the IBX employed in their previous report. I wonder if this is the result of simple optimisation, or if there's a significant KIE for the oxidation as well, which made them change conditions.
Anyway, the group then converted the aminoketone to N-methylwelwitindolinone C isothiocyanate as before and then onwards to a new natural product, N-methylwelwitindolinone C isonitrile, using Rawal's protocol with Mukaiyama's oxazaphospholidine. However, problems isolating the product lead to the development of a second route. This relied on Burgess dehydration of the formamide resulting from treatment of the amine with (in situ generated) acetic formic anhydride. Oxidation of N-methylwelwitindolinone C isothiocyanate and N-methylwelwitindolinone C isonitrile by treatment with base and air gave two of the oxidised welwitindolinones, whose structures were confirmed by x-ray studies and some computational work.
I'm sure a lot of people will just consider this an addendum to Garg's first paper, and I don't doubt it'll be decried by some as unJACSworthy, but I still think the chemistry is first rate. Great stuff!
References, Addenda and Slightly Irrelevant Ramblings
1. The whole back-to-back natural products syntheses isn't as rare as one might imagine, even in such a multidisciplinary journal as JACS. I blogged the Reddy/Porco Kibdelone C pair from last August, and See Arr Oh covered the Movassaghi/Tambar synthesis of trigonoliimines A-C in a delightful guest post over at Tot. Syn. a few months later.
2. I’d only be guessing, but I think the reason for the slightly more exotic electrophile (rather than, say, methyl formate, which is basically free) is the low reactivity of the copper enolate.
3. Although the aldehyde group obviously required some kind of alteration in reactivity to prevent it reacting with the hydrazine in place of the desired ketone Rawal did not originally intend to protect it by reduction. Initially, protection as the dioxolane was attempted, but upon heating an unusual Cope-isomerisation-aldol sequence occurred to deliver the product shown. I bet there was a bit of head scratching before the crystal structure was obtained.
4. Trace transition metal contamination can be a real pain. There's a good review on this in the current issue of Chem. Soc. Rev. I also plan to write something about this soon.
5. I'm not too sure how this works. Not being able to get the original papers doesn't help (our subscription to Tetrahedron and Tetradhedron Lett. only stretches back a decade). As a large part of the reason I write is to learn, here's what I reckon based on what Rawal says in the paper.
Suggestions? Also, as another aside, I see Aldrich sell this thiourea as a thiol (i.e. the other tautomer, 3,4,5,6-Tetrahydro-2-pyrimidinethiol). Rawal's first paper calls it propylenethiourea and draws it as I have, but the most recent one matches Aldrich. Anyone have any idea what the reality is? Certainly all of the thioureas I've made (none of which were cyclic, most of which were aryl) seemed to have a C=S IR absorption.
6. I'm quite sure I saw this or something damn similar used last week in place of the traditional phosphite additive in a Corey-Winter reaction, but I can't find the reference. Guess that makes sense as in the original paper Mukaiyama describes it as a phosphite replacement for the RNCS to RNC tranformation it's doing here.
7. But at what cost? Aldrich in the UK list Super Deuteride as discontinued, and I can't find any mention of it on Sciquest. Can anyone get a ball-park price?