Home » Still In The RBF

Oseltamivir phosphate (Tamiflu) Pt. 5   

29 November 2008 10,403 views 39 Comments

Shibasaki, Yamatsugu, Yin, Kamijo, Kimura and Kanai. ACIEE, 2008, EarlyView. DOI: 10.1002/anie.200804777. Article PDF Supporting Information Group Website

A fifth appearance for my favourite drug ‘interloper’ in to this natural-product space; alarm bells shouldn’t be ringing – just cause I work in pharma doesn’t mean I’ve turned my back on natural products! Tamiflu is of course based on a natural product, shikimic acid – the starting point for the original synthesis. But as natural sources go, it’s rather hard to get hold of, and thus damned pricey (£248 for 5g on SA just now). Other routes used involved chemistry that was perhaps a mite ‘tetchy’ on scale, such as azides and aziridines. A few years back, Corey announced that the synthesis was solved (which made the national press!!!), with his cracking synthesis that I still love from a chemist’s point of view. However, from a practicle perspective, there were still a few problems, including a -78 degrees or two, which are problematic on a plant scale.

On the same day his former student Masakatsu Shibasaki published his work on Tamifly, which was also a very nice synthesis, but the azides and aziridines were still rocking around the cyclhexane. It appears that he doesn’t think of them as a problem, as this latest synthesis is still loaded with those frisky nitrogenous beasties, but there’s loads more to this synthesis.

It kicks-off with an awesome asymmetric Diels Alder, in which they scoured the top-row of the periodic table for the right addive to induce the asymmetry. Barium isopropoxide did the job, along with a pretty damned complex ligand. A quick look on SciFinder tells us that it’s a five step synthesis, using some pretty nice chemistry in itself! It looks like it’s worth it, too – the result of the DA is very nice, building that asymmetry into the cyclcohexene with apparent ease.

A few steps later – and it’s azide time. Not just one acyl-azide, though; two in this case, all set to do their Curitus rearrangement to impressively build a cyclic carbamate and the Boc-protected amine in one pot. Nice work, as this also differentiates the two amines. However, they were aware of the problems of working with such a tempramental substrate, so they optimised the reaction conditions so the azides were never isolated or removed from their solvent.

After acetylation, they needed to bolt on a carbon, and preffered to use a C1 acyl anion equivalent – in this case, a protected hydroxy malononitrile. An allylic, Tsuji-Trost style opening and decarboxylation of the carbamate did the job nicely, installing the new stereocenter in apparently complete control (not that it matters in this case).

Completeion of the synthesis was a little traumatic – substrate controlled epoxidaiton went well, as did opening of the epoxide. However, the planned displacement of the newly installed hydroxyl to install the 3-pentyl ether was very difficult. They finally surmounted this problem by displacing it with the neighbouring amine (a double Mitsunobu), forming an azidiridine. Opening of this aziridine with Lewis acid and 3-pentanol finally finished the job (after salt fomation).

Nice work, using some impressive chemistry; read the paper for full discussion of all that Diels-Alder work!

1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)


  • gilgerto says:

    You might want to take a look at the synthesis of bryostatin 16 by Trost in Nature…

  • InfMP says:

    agreeed. this is a big one. intros by trost and then another summary in the issue by charette/.

  • tyrosine says:

    What do you have against azides and aziridines? These are kinetically stable functional groups that in the right hands, can be scaled up nicely.

  • milkshake says:

    Depends on the azide: higher MW alkyl azides are quite stable (or even extremely stable – if they have no alpha proton like trityl azide), aryl azides are much less so (with decomp onset temperature in the neighborhood of 100C) and acyl azides are touchy.

  • Tot. Syn. says:

    And of course the synthesis of Tamiflu used on scale by Roche was apparently limited by the tetchiness of the azide / aziridine steps, so they’re obviously a big deal in this case. I agree that they’re often not a problem, but sometimes…

  • Gilgerto says:

    By the way the double Curtius is quite nice, allowing the differentiation of both acyl azide groups in a single step. Well done.

  • vik says:

    Ya..the double curtius is cool ! Nice synthesis..but cant say about its commercial prospects. I always thought that a more practical way of tackling the Tamiflu shortage(2006) was by sysnthesis of quinic or shikimic acid followed by the current indutrial process. What do you think guys?

  • Gilgerto says:

    Anybody happy with the new ACS website… The previous site was just fine, the new one a piece of sh…

  • ch3mical says:

    The new ACS is better but the coverpages are annoying. Why should it matter when/where the article was downloaded? Is there anyway to prevent this?

  • Herr Lipp says:

    The new ACS website kinda blows….The ASAP emails are impossible to read (6pt font?).

  • anniechem says:

    There is one thing I LOVE about the new ACS website. If you download the ‘pdf w/ links’ version, all of the references are linked within the pdf. So, for example, you click on reference 8a and it actually brings you to the reference. It’s amazing. This only works, of course, if your institution has access, but so far for me, I’m thrilled.

  • milkshake says:

    Make sure to install the current Firefox version 3. Or better yet, download Chrome from Google, it is noticebly faster.

    The new ACS interface is awful with IE and Firefox 2 because of the graphical abstract aliasing and scaling problems.

  • dbstad says:

    My biggest complaint with the new ACS site is the fact that the RSS feeds (that worked beautifully before, with large graphical abstracts) are now updating a day later than usual, do not list full authorship and now have very small images for the abstracts. Anyone else have similar issues? I’m using NetNewsWire on a Mac currently. If there’s a fix I’ll gladly switch.

  • Conte says:

    you are right, the new ACS website doesn’t work as well as the old one, and I don’t see any reasons to change the old one, it was ok. anniechem, you are right, the links to the pdf are good, but was it necessary to change the website for this upgrade?

  • TWYI says:

    RSC websites >>>> ACS websites

  • ,,, says:

    And where did all the graphical abstracts go? Now you have to click on the thumbnails to quickly scroll through “what’s going on today”. Forget about “show all thumbnails”, it’s already slow enough!

  • q says:

    The ACS webpage used to be pretty good, but now the RSS feeds have become useless; the pdfs have annoying cover pages; the web pages look more cluttered and unproffesional; the graphics look terrible on IE. I don’t understand this “upgrade”.

  • Jose says:

    Well, since ACS membership and journal subscriptions are such bargains, budgets must be tight for web design, eh?

  • Tot. Syn. says:

    Y’know what? I really like the new ACS website, but then I’m into the web design thing, and it does look really nice. And I like the choice of PDF formats; the links work really well.

    However, the 5-image ‘highlights’ thing near the top of the journal page is pointless, along with a few other bits, like the ‘Announcements’, the ‘ACS Divisions’ bit, the ‘CEN Latest News’, and lastly, the tiny thumbnails in the abstracts…

    Perhaps they could have optional panels, like on the BBC website.

    Tot. Syn. gets it’s yearly facelift early next year – ideas anyone? Two of the previous designs can be seen here, on the WayBackMachine:



  • milkshake says:

    IE does not work well for the new look of ACS journals – you need Chrome or Firefox 3 (the new version) for the graphics to load properly.

    Google Chrome is also faster but unlike Firefox 3, it does not work well with ScienceDirect journal portals (such as Bioorg Med Chem Lett).

  • InfMP says:


    absolute crap. when i want the author’s most recent papers, I can’t do it that way anymore.

    The cover pages are disgracful. they don’t even look elegant (1 pixel per cm on that logo???). I have been deleting them manual after download

  • Undergrad senior says:



  • optional says:

    JACS on Facebook: next up: JACS on myspace.com and then on skype :)

  • Chiral pool's closed. says:

    Extrapolating further –

    LOLChemists: im in ur flahsk, settin ur stereochim… I can haz ee?

    JACS alerts on AIM: OMG!! Ru 4 EntSel CycIsom!!

    Whiny emo blog posts: Can you believe that hack over at {competing university} scooped me?? I’m totally unfriending him!!

  • PotStirrer says:

    New webpage is SOOOO slow. Arrrghh.

    @ Chiral pool’s closed
    Great name.
    Hilarious post.

  • cvengo says:

    one thing that is annoying about new JACS papers are those blue frames in which the schemes appear. Is there anything you can do with it? Sometimes, part of the molecule at the edge of the scheme is not visible.
    Otherwis I like the idea with links directly in pdf.

  • anniechem says:

    cvengo – you’re looking at ‘pdf w/ links’. The blue frames are there because if you click the image you can grab the image. Download the other version ‘high-res pdf’ and there is no frames, just the paper like usual, with the exception of the dumb cover page.

  • cvengo says:

    yeah, but then there are no links which I consider to be useful.
    I do not understand the idea with grabbing images, you can do so easily with acrobat.

  • milkshake says:

    Now that we are in full-blown bitching mode – does anybody know some easy-to-use fast reliable ACD search interface?

    Our institute lost the subscription for ACD/MDL-ISIS, a database of all commercially available chemicals and their prices – an exceptionally important database. Now we areall supposed to use ACD search through Symyx “Discovery Gate” instead. The damned thing is a Java-heavy, poorly designed and cluttered web page interface. Life is pain with Symyx..

  • anon says:

    milkshake – we’re stuck with Discoverygate as well, and I agree that it’s a vile POS. Can’t help you, but if you unearth a good alternative, please post it here…

  • sjb says:

    Re facelift.

    Whilst I appreciate the idea, why do you feel you need to have it done – if it ain’t broke don’t fix it :)

  • Jose says:

    milkshake- chemspider might be helpful.

  • TB says:

    Shikimic acid is made very cheaply using a recombinant E. coli strain. This is not a limitation at all. See the article below.

    Shikimic acid and the synthesis of Tamiflu. Frost’s US 6,613,552. The New York Times quoted Ernie Prisbe, vice president for chemical development at Gilead Sciences (which holds the patent for Tamiflu): Whether for Gilead or for others, when it comes to making Tamiflu “on a scale of 1 to 10 in difficulty, this is maybe a 7 or so. There’s nothing that overwhelming in this kind of synthesis, or that formidable, that someone couldn’t do it.”

    The synthesis of Tamiflu can start with shikimic acid (although it does not have to). Shikimic acid can be extracted from star anise.

    Of the allegations that limited supplies of star anise are a bottleneck to making Tamiflu, Professor John W. Frost of Michigan State University states: “I’m just completely astonished about the gnashing of teeth and the wringing of hands about the shikimic acid. The bottleneck should not be shikimic acid availability.”

    Frost developed, and patented, and method for making the starting material, shikimic acid, without utilizing star anise. Roche has used Professor Frost’s method in recent years, but Frost says he heard the company had cut back. In the fiscal year that ended in June, 2005, Michigan State received $113,000 in royalties from Roche, according to Paul Hunt, the university’s associate vice president for research and graduate studies. Roche pays $1.85 for each kilogram of shikimic acid it makes using the process, indicating it made about 60 metric tons in that year.

    Frost’s patent, titled Biocatalytic synthesis of shikimic acid, is US 6,613,552, issued September 2, 2003. The patent notes: Work on this invention was sponsored in part by the United States Department
    Of Agriculture Grant No. 95-37500-1930 and the National Science
    Foundation Grant No. CHE963368 amendment 002. The Government may have certain rights in the invention.

  • PotStirrer says:

    Milkshake. You can make do with SciFinder by toggling “commercial availability” when you do a structure search, but I fondly remember the good old days when I had the MDL-ISIS access.

  • vik says:

    TB: Cool ! Didnt know that. I did read Frost’s paper in JACS 2001.. i think and remember reading that the process needs significant developments. But that was 7 years ago.

  • InfMP says:

    the commercial availability option sucks now.

    it only gives prices for a few companies.

  • Jose says:

    The SciFinder comm avail thing is bogus- they upload whatever catalogs vendors send them, so it is wholly incomplete and lists many bogus compounds that don’t actually exist -a Russian and a French company are notorious for listing things they don’t actually have.

  • RBW says:

    All these gripes about the ACS webpage! Well, when I was young the Internet did not exist, and we had to patiently wait for every journal to arrive by mail (often across the oceans…).
    Back to chemistry-
    1) It is unlikely any academic synthesis of Tamiflu would be considered as a replacement for the industrial route unless it is far far superior. It would just be too expensive to switch to a new route otherwise.
    2) The starting Diels-Alder could easily be achived by auxiliaries. For example, the bis-menthol ester of fumarate gives great enantioselectivity as was known many years ago. Menthol is cheap, so is it really progress to use a CAPO catalyst that takes 5 steps to make? Debate at will.