Here's an impressive total synthesis of schindilactone A by Tang, Chen, Yang, and 14 coworkers. At 29 steps in the longest linear sequence that's comfortably fewer than two per author. Still, the route is entirely linear and it's a fairly heroic effort, as we'll see.
Work began sometime ago as the group published syntheses of the ABC (Org. Lett., 2006, 8, 107) and FGH (Org. Lett., 2005, 7,885) fragments of the slightly more complex micrandilactone A some time ago. Apparently that unique, ketal spanned, 7-8 carbocyclic system in the middle took some time to work out. I'll cover the older work on those fragments as well, as it shows the origins of some of the key steps in the schindilactone A total synthesis.
The route to the ABC chunk of micrandilactone A began with my favourite named reaction: the Diels-Alder. It turns out that the regioisomer they get is by far the major product, but in the absence of a Lewis acid the regioselectivity is reversed. This surprises me as the TiCl4 should presumably complex the more Lewis basic ester in preference to the ketone, which would favour the isomer that isn't formed. And it's not coordinating the oxygen in the diene because the reaction also gives this regiochemistry if the diene is simply isoprene. Anyway, the Diels-Alder gave the expected trans-6-5 system, which underwent epimerisation during the Grignard addition - lactonisation step. The lactone was then a-hydroxylated in the most economical way possible; using base and oxygen. This step is performed here on 3g, and later on 10g (for schindilactone A), a scale on which MoOPH, oxaziridines, or even the Rubottom oxidation aren't too tempting. Reduction of the lactone to the lactol with lithium aluminium hydride set the stage for a rather clever anulation reaction using triethylphosphonoacetate to install the second ring. The cyclohexene was then converted to the ketoaldehyde by ozonolysis, quenching with dimethyl sulfide. The aldehyde was then selectively reduced to the alcohol using freshly prepared Raney Nickel under hydrogen, which didn't touch the ketone (or the benzyl ether). This alcohol was protected as the TBS ether and then treated with the organocerium reagent derived from lithium trimethylsilylacetylide. Strangely the paper text says that catalytic CeCl3 was used when the SI says that a whole equivalent was employed, and even mentions forming the organocerium and then adding this to the ketone. The newly formed propargylic alcohol was then protected as its acetate ester, which required heating for a fairly long time in toluene, presumably due to its hindered nature. Deprotection, oxidation, and methylenation of the TBS-protected alcohol gave the eneyne precursor, which, using a fairly generous spoonful of Grubbs 2nd generation catalyst, closed to give the ABC ring system.
The group's previous preparation of the micrandilactone A FGH system starts with another reaction which I like, but have never had an excuse to perform - the Pauson-Khand cyclopentenone synthesis. For me this always brings back memories of seeing P. Andrew Evans talk, and how his accent is different every time. Here, the reaction worked in excellent (80%) yield using standard conditions, with superstoichiometic Co2(CO)8, but this method was deemed too impractical and expensive for such an early step, so a catalytic process was developed. It was found that the reaction was greatly improved by the addition of tetramethylthiourea, the use of which as an additive had been previously reported by the group, and just 5 mol% Co2(CO)8 was needed. Luche reduction and protection of the enone, followed by reduction (and reoxidation) of the lactone gave the g-hydroxyaldehyde. This was reacted with a threefold excess of vinyl magnesium bromide to give a 3:2 mixture of epimeric alcohols in favour of the desired compound. Although this compound did undergo the key Pd−thiourea-catalysed carbonylative annulation, the yields were poor and some interesting byproducts formed (see the paper if you like palladium chemistry). None of the desired product was formed in the absence of the thiourea ligand. Fortunately, it was found that if epoxidation of the tetra-substituted olefin was performed before the annulation step, rather than after, the yield increased to essentially quantitative. A few standard transformations gave the FGH enone, although apparently with the wrong diol stereochemistry for the natural product.
Okay, so now we're all up to speed, let's take a look at what was published last week! Don't forget that we talking about a different target now, so the fragments above don't quite fit, although the chemistry's very similar. For schindilactone A a completely new, and quite cool synthesis of the BC rings was devised. Next came the D and E rings, and followed by the F, G and H rings, which were established using pretty much the same chemistry we've just seen. The last ring to be constructed was the A ring. Unfortunately, it's all a bit linear, but I guess 29 steps isn't too bad if you've got 14 eager coworkers.
The route started much as before, with a Diels-Alder reaction, Grignard addition and a-hydroxylation with oxygen. The new hydroxyl was protected as its TES ether, and then dibromocarbene was used perform a cyclopropanation of the silyl enol ether. Treatment of the dibromocyclopropane with silver perchlorate then caused the desired ring expansion in good yield to give the 5-7 BC system. A palladium mediated cross coupling between the vinyl bromide and the silyl ketene acetal gave the b,g-unsaturated ester in excellent yield. Grignard addition with but-3-enylmagnesium bromide to the ketone then occurred with impressive diastereoselectivity, attributed to the bulk of the nearby TES ether, and the newly formed alcohol cyclised onto the nearby ester. Although some double bond isomerisation did occur, this didn't matter as the mixture could be deprotonated with KHMDS to give a single dienolate which reacted selectively in the a-position with MoOPH. The new hydroxy was protected as its benzyl ether and the lactone was then treated with vinyl magnesium bromide. RCM using Grubbs' second generation catalyst in the presence of magnesium bromide successfully formed the challenging 8 membered ring and epimerised the hemiketal to the required configuration. The F ring was then installed by conversion of the alcohol to its tetrolate followed by Pauson-Khand reaction as before.
This hexacyclic compound was then further functionalised to get to a suitable precursor for the key carbonylation reaction. Thus, the diol below was prepared, and did undergo the expected reaction, although a fairly high loading of catalyst and ligand was required. This lactone was then a-methylated, which gave the wrong diastereomer as the major product. Fortunately, deprotonation with LiTMP and then quenching with ammonium chloride solution was able to effect epimerisation to the correct product. Apparently, both methylation and protonation of the enolate occur from the bottom face. Finally, the TES ether was converted to the acetate ester, the benzyl protected group on the D-ring hydroxyl was cleaved, and the acetate was then cyclised onto the nearby lactone by treatment with LiHMDS. Dess-Martin oxidation of the allyl alcohol gave the natural product.
Apparently efforts towards an enantioselective version are ongoing. Good luck!
1. You may remember controversial first total synthesis of maoecrystal V by Yang (J. Am. Chem. Soc., 2010, 132, 16745-16746; early work in Org. Lett., 2009, 11, 4770-4773) from last year, which famously appeared to have a great deal in common with an earlier approach by Baran (Org. Lett., 2009, 11, 4474). Oh, and a former Baran postdoc was on the Yang paper. For some lively discussion see: http://totallysynthetic.com/blog/?p=2577. Aside from this, Yang's actually quite the synthetic chemist, with some interesting work on the cortistatins and pseudolaric Acid A. His website says he's also working on Solanoeclepin A. Yikes.
2. I found out just last week that there's an HMPA free edition, using everyone's favourite phosphoramide surrogate, DMPU. MoOPD, a Jim Anderson product, can be found in Synlett, 1990, 107-108.
3. I hear that Lewis bases, amine oxides (NMO and TMAO), or even photolysis can be used to accelerate the reaction by helping to open a free site for the alkene to coordinate, increasing the rate of turnover.
4. Remember - enones generally deprotonate g but react a.