Totally Synthetic by Paul H. Docherty, 6 May 2008
Total Synthesis of 35-Deoxy Amphotericin B Methyl Ester
A. M. Szpilman, D. M. Cereghetti, N. R. Wurtz, J. M. Manthorpe, E. M. Carreira, Angew. Chem. Int. Ed. 2008, 47, 4335-4338.
A. M. Szpilman, J. M. Manthorpe, E. M. Carreira, Angew. Chem. Int. Ed. 2008, 47, 4339-4342.
35-Deoxy Amphotericin B Methyl Ester is not really a natural product, but close enough. It is actually the well-known amphotericin B with a missing hydroxyl group designed by Carreira to probe some biological activity. In addition, he has developed a completely new route to the natural product class. A previous total synthesis of the parent compound was published by KC Nicolaou in the late eighties, so it’s interesting to see what’s changed in two decades…
Time for a retroanalysis:
The synthesis of the septaene (new word!) fragment is actually less interesting that one might imagine (though I reckon it’s probably rather frisky in the RBF…). More involved is that every-pesky glycosidation step, where model studies seem not to be very useful. A nitrile-oxide addition provided one of the key C-C bond formations, whilst Carreira’s own acetylene chemistry dealt with another. Let’s start with that:
So the starting material for these substrates is enantiomerically enrichate
dialkyl malate; (S) for the top fragments, (R) for the bottom. A
Prasad reduction provided the desired syn-1,3-diol relationship in both
cases, whilst a nice
Ohira-Bestmann reaction installed the required acetylene.
The reaction with the lithium acetylide gave undesired stereochemistry in the
addition step, but an asymmetric zinc catalyst connected the terminal alkyne to
Conveniently, Carreira has
been working on
that chemistry for a few years. And a nice result it is too - high
yield and d.r.
The next reaction is one of my favourite macrolide-fragment-coupling reactions. Taking the product of the latter reaction and converting it swiftly to the oxime allowed them to do a nitrile-oxide cycloaddition (a [3+2] or 1,3-dipolar reaction) onto the terminal olefin to give the 4,5-dihydroisoxazole product.
The really smart bit here is the cleavage of the N-O bond; some molybdenum hexacarbonyl liberated a hydroxy ketone, which cyclised during purification to give the desired product. A very interesting reaction!
The synthesis of the other large fragment isn’t discussed in detail in the paper, but the stereochemistry of the methyl group is set using Fráter-Seebach alkylation, followed by a Myers alkylation. After construction of the polyunsaturated chain, the linear molecule was completed by Yamaguchi esterification, leaving a HWE to complete the macrocycle.
As in many cases, glycosidation was a process of trial and error. That’s not to say that they didn’t think hard about what might work, and what was causing them trouble - this chemistry is just hard. After a lot of work, they found that the substrate below was the best partner for the algycon, and that activation with 2-chloro-6-methyl-pyridinium triflate and 2-chloro-6-methyl-pyridine resulted in a 45% yield of the desired product.
Deprotection and reduction of the azide provided the target, which turned out to be an order of magnitude less active than the parent compound. This, however, backed up their biological hypothesis, which I suggest you read about yourself in these two impressive papers. Also, congratulations to Alex for his work on both these articles - good job, mate!
First of all, nice work and congratulations on a tremendous accomplishment. I noticed that you guys are big fans of the TEMPO oxidation of primary alcohols. Is there a reason why you prefer that oxidation in particular when there are so many other options available?
Yes, there are several reasons
1. All the reagents are commercially available (you do need to titrate the bleach to make sure not to add an excess).
2. It is easily scalable. On very large scale some strong stirring may be required to keep the temperature down.
3. It is fast (usually done in a few minutes).
The aldehyde is usually clean enough after water workup for use without any flash (e.g. for use in the Zinc acetylide addition or Ohira reaction.
4. Most importantly the yields are high. The yields usually only drop if you add the bleach to slowly or if the temperature runs away with you.
5. When necessary it is selective for primary alcohols over secondary alcohols.
The bleach titration is with starch/iodine?
yeah, DMSO from Swern or NMO from TPAP oxidation would probably ruin the enantioselectivity of the acetylene addition.
By the way, the alcohol corresponding to the aldehyde depicted here (in paper the compound 5) is a side-chain building block for Lipitor and Crestor statins. It is commercially available from Takasago – we bought 1 kilo bottle of the stuff for a project that is over now and this optically pure acetonide-ester-alcohol is gathering dust in the cabinet…
… yes the bleach titration uses starch, KI, and sodium thiosulfate. The bleach oxidizes the KI to I2 and then you use the thiosulfate to titrate the I2. Here’s a procedure:
As you note, the installation of the mycosamine sugar unit was particularly nasty. Jeff and I had days and days of fun solving that problem.
The 2-chloro-butytic ester was the only ester that at the same time gave more of the glycoside than the orthoester, was able to control the formation of the desired beta anomer and could be removed under mild basic conditions (thank you chlorine).
The use of the mild and sterically hindered acid 2-chloro-6-methyl-pyridinium triflate was born out of the formation of adducts between the sugar donor and the PPTS we originally applied (Nicolaou’s procedure) as described in the paper. Strong acids would rip out the unsaturated alcohol on the aglycone leading to the formation of the worlds longest allylic cation (strongly blue). Weaker ones would not induce reaction.