If you're reading this then I reckon you've probably heard of taxol, as it's one of the most talked about synthetic targets of all time. It's a molecule with a fascinating history, from its isolation and structural assignment in 1971, to discovery of its potent activity and interesting mode of action, and the ensuing scrabble to solve the supply problems plaguing its development as a drug. Its rise to success as a billion dollar pharmaceutical was stellar, and is one of my favourite examples of how useful and important synthetic organic chemistry is. Although it's been 5 years since the most recently reported synthesis of taxol, last week's Baran synthesis of taxadiene (following his cyclase-oxidase mimic plan for elaboration of this hydrocarbon into taxol) seems to have again gotten chemists talking excitedly about this target. After overhearing things like 'didn't Nicolaou make it first?', 'there've only been x syntheses so far' (where x is 0-6) and other such misinformation in our office I've decided to take action. Yes, there are already numerous reviews, book chapters and even entire books on this subject, but it seems that a lot of people don't have the time or inclination to read them. So, here's a brief summary of the 7 syntheses published so far, in the order they were completed. Hopefully this'll help put recent developments in perspective.
The actual first synthesis of taxol, albeit it as the unnatural antipode. Counting the number of steps for this route was particularly difficult as the origin of the starting material isn't obvious from reading the original papers. Although it's described as 'readily available', as far as I can tell no procedure for its preparation is given. An previously reported literature preparation by Büchi took 16 steps, and if this route was used then Holton could be a strong contender for the longest synthesis on record. Additionally, natural inexpensive (£0.26/g) (+)-camphor actually gave the wrong enantiomer of the natural product. In principle, (-)-camphor could be used to give the correct enantiomer, although it is far more costly (£10/g). I love the Chan rearrangement, which is the closest thing we're likely to get to a retro-Baeyer-Villager for some time... would you include one in your synthetic plan?
Nature 1994, 367, 630 [PDF]
J. Am. Chem. Soc. 1994, 116, 591 [PDF]
Often referred to (by KCN) as the 'first published' synthesis of taxol, which it was. However, this is a bit misleading as Holton did complete and submit his synthesis for publication over a month before the Nicolaou group, but the JACS editorial office took two months to get it into print, making it the second of the two to appear in the literature. Having said this, Nicolaou's synthesis is a great deal more efficient than Holton's, and is convergent, disconnecting the natural product through the central B-ring. The formation of this challenging medium size ring by an intramolecular McMurry-type coupling is undoubtedly one of the highlights of the route.
J. Am. Chem. Soc., 1996, 118, 2843 [PDF]
Although as couple of years after the first reported syntheses, and quite a bit longer than Nicolaou's route, Danishefsky's solution to the taxol problem features some very interesting chemistry to overcome unexpected obstacles. From the paper:
"The most rewarding aspect of the synthesis was the ability to start with Wieland-Miescher ketone, itself available through catalytic asymmetric induction, and to install all of the stereochemistry required to reach baccatin III in a sequential fashion. Our synthesis, though arduous, involves no relays, no resolutions, and no recourse to awkwardly available antipodes of the “chiral pool”."
Danishefsky's route also differs from the others in its markedly different order of ring construction, which involves first establishing the CD rings and then adding the other two, carrying the oxetane through a large number of synthetic operations.
Conceptually, Wender's route is more similar to Holton's than that of Nicolaou, drawing its asymmetry from the chiral pool by starting from a simpler natural product, and also relying on the fragmentation of an epoxyalcohol to form the 8-membered B-ring. Despite its linear nature the route is highly creative, uses an inexpensive natural starting material, and at only around 37 steps in length is the shortest to date.
Another convergent approach, but still bearing an awkward and lengthy longest linear sequence. Unlike the other routes, which either dip into the chiral pool directly or use resolution, here an enantioselective route is achieved through the use of a Sharpless Asymmetric Epoxidation reaction. There are several cool steps, including reductive rupture of a cyclopropane ring to introduce a hindered methyl group.
Chem. Eur. J., 1999, 5, 121 [PDF]
Another fairly lengthy route from Japan, made interesting by the unusual tactic of constructing the B-ring cyclooctanone first and then introducing the smaller rings sequentially.
7. Takahashi (2006) [Wiki]
Chem. Asian J., 2006, 1, 370 – 383 [PDF]
Although only a formal (and racemic) synthesis of taxol this route does have the fewest authors of any Taxol total synthesis, thanks to a large part of the work being carried out using an 'automated synthesizer[sic]'. I'm extremely sceptical of the final stages of the route, where 5 successive steps are all carried out on less than 1 mg, and yet full data (including melting points, some of which are 'decomposition') are reported. Hmmm. However, on the plus side I did particularly enjoy the old school method for B-ring formation.
References and addenda
1. According to T. Hud's excellent chapter in The Way of Synthesis, at the height of taxol mania more than 30 groups were actively pursuing its synthesis.
2. For posterity: Holton's manuscript was received by the JACS editors on the 21st of December 1993 but not printed until the 4th issue of the year on the 23rd of February 1994; Nicolaou's wasn't received at the Nature offices until the 24th of January but appeared in print some five days before Holton's on 17th of February.
3. Cyclopropanes are also a great way to introduce gem-dimethyl groups next to carbonyls, a transformation that's not easy in hindered or cyclic substrates. Simply alkylate with 1,2-dibromoethane and then use hydrogenolysis to break the cyclopropane open to give two methyl groups. Sweet.
4. Is this the future? If you were wondering:
5. This is a seriously retro ring closure, developed by Stork in the 1970s and employed in his classic PGF2α synthesis way back in 1978.