Our paper is now out. This ongoing project, although some might label as simple, requires many different types of oxidations, and all of which require many experiments to discover. You can read the paper and ask me questions or send me comments to your heart’s content, but in this post I will talk about the discovery process and tell some of the stories.
Here is a graphical summary of the paper in one hilariously hideous scheme (No, Phil, I don’t think it’s hideous just because the fonts are in Arial):
We found the first oxidation (at C-5 of taxadiene (1)) in palladium-catalyzed acetoxylation (treat your olefin with Pd(OAc)2 and benzoquinone in acetic acid). This is one of those reactions that looks very clean by TLC and crude NMR, but you only get 35% yield in the end. I wanted to see if I could figure out where the starting material was going, so I ran the reaction in an NMR tube with deuterated acetic acid and 1,3,5-trimethoxybenzene (TMB) as an internal standard as the only deviations from the original experiment. I came up with these data below:
I still couldn’t see any taxanes other than the starting material and product, which was kind of frustrating, but I did notice this reaction gave 42% yield with some SM left rather than 35% with no SM left. After corroborating this result with the isolated yield of a separate larger-scale reaction, we concluded TMB was somehow giving us the increased yield. A quick screen of similar molecules led us to find that using only 4 equivalents of anisole gives us a 49% yield. Like we describe briefly in the paper, we qualitatively observe much less palladium black forming in these reactions containing anisole, so maybe anisole facilitates reoxidation of Pd(0). This is a somewhat simplistic explanation because we saw 35% yield even when using stoichiometric Pd(OAc)2, so maybe palladium black actually hinders the continued reaction. People smarter than me have suggested anisole improves the yield because it is a “redox buffer” or a “cation sponge,” but I don’t have any evidence for those ideas.
The next oxidation at C-13 was pretty extensively covered in the paper, so I will be somewhat quick about it. We were getting byproducts 7 and 8 from Cr(VI) reagents like PCC and CrO3∙3,5-dimethylpyrazole complex (see Scheme 2 in the paper for a clearer view of this stuff, but you can see it above too). I was in my first year as a grad student, and was trying as many different Cr(VI) reagents we could think of, both known and unknown (you can see a bit of my interesting screen near the bottom of the supporting information). While I was putting my head down and working hard to find a solution, Abraham came to me and said he was going to try this Cr(V) reagent (9) he found in the literature. Apparently it was shown to be less effective at oxidative cleavage of 1,2-diols than Cr(VI) reagents. As the naïve student lacking the courage it takes to try something unknown that I was, I told him he was crazy. This Cr(V) reagent gave us an improved yield and a totally different byproduct. For a long time we couldn’t figure out what to do with this reagent to determine its capabilities, but after thinking about the mechanism for the byproduct’s formation we decided to see if it does the allylic transposition of the Babler-Dauben reaction. It doesn’t (see bottom rectangle in the above messy scheme), which we think is pretty interesting. Oh and I should note that we saw slightly lower yields with acetonitrile instead of trifluorotoluene as solvent, but with acetonitrile we didn't need to use 15-crown-5. So, for simplicity's sake, if you want to try this reagent on something new I would suggest using acetonitrile and no crown ether.
The third oxidation, perhaps unsurprisingly, comes with a story too. Abraham and I tried quite a few types of oxidation with C-13 at either the ketone or the alcohol oxidation state (corresponding to compounds 5 and 6 in the paper, respectively; see Figure 1 in the paper or the horrible scheme above). We neglected one type of allylic oxidation, and it took a brand new postdoc (Minetaka) to show us that simple radical halogenation—that everyone learns in first-year undergraduate organic chemistry—was very selective for C-10 oxidation. I find it super interesting that excess NBS brominates C-18 (the neighboring vinyl methyl group) for enone 5, but treating diacetate 6 first brominates this C-18 methyl and then C-10. This experiment is actually corroborated by computationally modelling these respective allylic radical species, which is kind of cool too.
After discovering the first three allylic oxidations of our synthetic oxidase phase, we devised a way to get to (–)-taxuyunnanine D (3). This took time because we wanted to find a way to do it concisely because that’s sort of our thing. I think it turned out pretty well.
What did I learn from these years of oxidative explorations? Pay attention and maybe you’ll find something you weren’t expecting. Also, be brave and try new and crazy things! Finally, spreading out the work among the team, or even just discussing your work with others, will often lead to solving long-standing problems.
I think this goes without saying for everyone else in the lab too, but I am totally willing to post or send my original FID files if you think I doctored my spectra or if you need them for some reason. This is the Open Flask after all.
The current version of the paper online has mistakes, most notably the missing and cutoff words in Figure 1. It will be fixed soon. Edit (20 March 2014): Figure 1 is legible now, but some arrows are very pixelated. Still working on it.