Thursday, August 1, 2013

Ingenol: Behind the scenes


Disappointment, frustration, excitement, setbacks, thrill, and success. It has been a roller-coaster ride, but finally the reward is here. A total synthesis of ingenol from our lab was published online today. The chemistry is all described in the paper, so I will not go into too much detail here. However, I want to highlight some of the key reactions and hurdles we had to overcome to finally being able to publish this work.
Our synthesis starts from commercially available and cheap (+)-3-carene, and in five steps we get to the substrate for the first key reaction, the Pauson-Khand. In order to get enough of the Pauson-Khand product, we needed to optimize this reaction quite a bit, but we can run it on gram-scale now as shown below.


Gram-scale Pauson-Khand reaction

1,2-Addition of MeMgBr to the carbonyl group of the P-K product gives us the so-called cyclase phase end-point.

Happy Christian with 1 gram of cyclase-phase endpoint

The oxidase phase also consists of seven steps, in which we install the four hydroxyl groups and rearrange the tigliane skeleton to give the desired ingenane skeleton.
The most troublesome, frustrating, and time-consuming problem to overcome was the pinacol shift to give the ingenane skeleton.


After 8 months of desperation, numerous shift reactions in one form or the other, a lot of disappointing results, and a ridiculous amount of NMR time to figure out what we got from all these reactions, we still couldn’t get the desired product. In fact, we actually almost abandoned the whole route because we just couldn’t get that shift to work the way we wanted. Eventually, we agreed to start designing another route to ingenol, if we could not get the shift to work by November 1st, 2012.
You may call it insane luck, but on October 24th 2012 we finally hit the sweet spot and got the reaction to work on a model substrate and about a week later on the real system. What a relief! Finally we could move on.
Comparably, the last steps of the synthesis were a much smoother ride. An allylic oxidation, a deprotection, and an elimination step later we eventually had a tiny amount of our first natural product, 20-deoxyingenol, just before Christmas of 2012. About a month later, we finally made ingenol for the first time.
However, it took us another 6 months of optimization and scale-up to get the end result, which was published online on Science Express today.
Especially, the reductive alkylation of the chloro-ketone needed some serious optimization. Initially, we tried to prepare big quantities of the methyl ketone itself which turned out to be very difficult to prepare and handle. We had some really painful weeks where we would run this reaction almost everyday and get out nearly nothing. At some point, Phil suggested to run both steps - reductive methylation and aldol reaction - in one pot (!) without isolating the methyl ketone itself. Initially we had a hard time imagining that this would work but after some optimization we finally arrived at the one-pot methylation aldol procedure.


Frustration after another failed attempt at the reductive methylation

What do you do if you almost go nuts? You do stuff like drawing your team-mates.
Here is Christian’s artistic interpretation of Steve and myself.

So what’s up next? We are currently collecting data for an upcoming full paper, which will cover the side reactions and failed approaches. Furthermore, we are focusing on making analogs of ingenol for pharmacological testing. The key intermediate, the cyclase phase end-point, which is our branch point for analog syntheses is being scaled–up in conjunction with our collaborators at LEO Pharma, developers of Picato® (an ester of ingenol).


Whoever is reading this post, if any, please feel free to comment and ask us questions about the synthesis. We will be more than happy to answer them and give you more insights.

Cheers

25 comments:

  1. What are your current thoughts for replacing the "toxic, stoichiometric" late-stage oxidants?

    ReplyDelete
  2. Yeah, I was wondering that as well. Beyond the OsO4 (which gets a shoutout in the paper), how much of the late stage oxidation chemistry was driven by necessity? (ie. nothing else seemed to do the trick)

    ReplyDelete
  3. Awesome, JUST AWESOME. Congratulations!!!

    One question: for the key step, isn't BF3 etherate getting quite commonly used in pinacol rearrangment? I know it's hard, just asking...

    ReplyDelete
    Replies
    1. Thank you! It turns out that the particular reaction conditions (temperature & solvent) as well as the carbonate protecting group were a more key find (although BF3 is the only lewis acid we've ever seen work). If you don't do the reaction just right you won't see any of the desired product or any trace that it was ever there. We're working on a full paper where we'll have a lot of gory detail on the particulars of that reaction.

      Delete
  4. I find this really interesting, but I'm not at all experienced in this field, so I'm not sure if the following questions are lame. Please excuse me, if that's the case:

    1. What's the role of DMAP in the first step?
    2. Many of your columns have been packed with DCM, but eluted with a totally different solvent. Why?
    3. After you've prepared the aldehyde with IBX, how did you determine the concentration in the solution or did you just assume that the reaction went quantitatively?

    ReplyDelete
  5. @See Arr Oh and Brandon Findlay:
    For the OsO4 we gave a clue in the paper as to how we will solve it. Catalytic dihydroxylation of the substrate in the current route was not successful for us. For the Selenium we only scratched the surface in terms of reagent selection. SeO2 was one of the first things we tried and it worked pretty well, so we did not look much more into that. There are ways of rendering Se catalytic.
    Ingenol is not the API, so even if there are slight toxic impurities that's OK. The ingenol synthesis will be a non-GMP synthesis as there are three additional GMP steps with crystalline material to produce ingenol mebutate (or analogs thereof).
    @Lee Kim: Yes, BF3 etherate seems almost too obvious when you look at it now. But it's not just the reagent that made this reaction work. The combination of solvent, temperature, reagent, and how you quench the reaction, was crucial for the success of this step.

    ReplyDelete
  6. Congrats, team! I'll leave the finer points of the synthesis to others, but I wanted to get your side of a cool story the folks at LEO told me during my interview. Apparently the lab regularly shared your presentation slides with the LEO team in advance of teleconferences. And then a little bit after you had the October relief, as you call it, there was a meeting coming up, and you DIDN'T SHARE THE SLIDES AHEAD OF TIME. Apparently Phil also told the LEO team, "Oh, you folks should really be on this call, there's some stuff I really need to go over." I think that was when the LEO folks had their first hunch that you'd finished. How close is my recollection to how it really went down? To what extent had you planned the "reveal" out on purpose? - Carmen (C&EN)

    ReplyDelete
    Replies
    1. I think your recollection to how it went down is pretty close to what actually happened. I'm not sure exactly what Phil told the LEO team before that TC, but they knew something good was coming, I think. We wanted it to be a nice surprise for the LEO team, it was Christmas time after all :)

      Delete
  7. 3 comments: 1) Kudos for including actual photographs of the reaction setups in the experimental supplementary. Extremely helpful. I hope more people will do do this. 2) You seem to be using smooth 24/40 septa for vacuum work- we had noticed that (depending on the manufacturer) sometimes the large 24/40 septa do not hold vacuum nearly as good so we prefer serrated SubaSeal white septa for highvac applications 3) If you need to replace HMPA at some later point, in the Li-naphthalenide reduction/methylation, using a combo of stoechiometric 18-crown-6 and potassium naphthalenide might do the trick. (Crown is best dried by azeotroping it with toluene).

    ReplyDelete
    Replies
    1. Thank you very much for your comments and nice suggestions. And I agree, these septa are not very good for highvac applications, however, for removal of the methyl iodide we only go down to ca. 200 torr.

      Delete
    2. Its not just about getting good vacuum - more concern is the oxygen and the moisture that gets sucked in. (With 0.01 Torr vacuum the pressure differential against atmosphere is about 760 Torr. At 200 Torr the pressure differential against atmosphere is about 560 Torr.) The best sealing SubaSeal are from Fisher, catalog # FB-686-90, pack of 10 costs about $30. (Aldrich similar Suba septa product is not nearly as air-tight)

      Delete
    3. Ok, thank you for the info. To be honest, I'm not even sure which brand of septa we use or how much they cost. But I will definitely keep your advise in mind, if I need a good seal in the future. Thx.

      Delete
  8. Did you consider 4 + 2 cycloaddition for the construction of seven membered ring ?

    ReplyDelete
    Replies
    1. No, a 4+2 was never really part of the plan.

      Delete
  9. @Noobchem: let me try to answer your questions.

    1) Our original conditions for the chlorination were different from what we actually used in the end. They worked too but were more painful to run and not suitable for scale-up. When we optimized this step, we at some point found that NCS effected the chlorination but the reaction was very slow and did not go to completion. However, when we added DMAP, the reaction was done in 3h. The way DMAP works in this reaction is probably similar to what they observed here: http://www.nature.com/nature/journal/v445/n7130/abs/nature05553.html

    2) No special reason, just personal prefence: Whenever my crude product does not dissolve in the eluent that I actually want to use (e.g. hex/EtOAc), I pack the column in DCM, dissolve the crude in DCM too and load it on the column and THEN elute with the actual solvent system I want. This works quite well.

    3) As for the concentration of the aldehyde: we determined the concentration by NMR using an internal standard.

    ReplyDelete
  10. Just a question Lars, how many chemists were working on this project and how long (months) did it take from start to finish ?

    ReplyDelete
    Replies
    1. By the time we finished the synthesis the project had been running for about two years followed by six months of optimization. 2-3 chemists have been working on the project at all times.

      Delete
    2. Making ingenol analogs for LEO Pharma - that part of work is still ongoing? Are you also trying pharma-process friendly optimizations (replacing HMPA, non-chromatographic purifications, etc.)?

      Delete
    3. Yes, analog work is still ongoing and so is optimization of the route. Scale-up is being outsourced in conjunction with LEO Pharma.

      Delete
    4. there is a company selling silica-supported Na-K liquid alloy in a formulation that is a gray-black non-pyrophoric free-flowing powder. I wonder if it would be worth trying, with sub-stoechiometric naphthalene and maybe some crown, for the Li-naphthalenide/MeI HMPA step

      Delete
    5. It's definitely a good idea as long as there's a catalytic turnover at the low temperature we need to do this reaction at. I'll look more into that. Thank you!

      Delete
  11. I was never brave enough to work with neat Na-K liquid alloy myself, hearing nasty stories about fires, but for the extremist chemists there is a Na-K-Cs ternary alloy that stays liquid even at -78C and reportedly reduces benzene to cesium benzenide radical anion...

    ReplyDelete
    Replies
    1. Yes, I usually try to avoid that stuff too. Our scale-up team doesn't complain too much about the Li-naphthalenide anyways, so I think our time would be better spend on trying to replace the HMPA with something else and optimize some of the later steps.

      Delete
  12. as long as you have Li-napthalenide and LiN(TMS)2 you need a lot of HMPA-like additive to improve the reactivity of Li enolate. But if you could switch to K-napthalenide and KN(TMS)2, 18-crown-6 might have similar effect...

    Et2NSO2NEt2 is a Grignard-compatible solvent with properties similar to tetraalkyl ureas (TMU, DMEU, DMPU). Also I would try amines: N-Me-morpholine, TMEDA and pentamethyl diethylenetriamine are useful for enhancing Li enolates/ Cheap and easy to get rid of.

    Also I was thinking, if you can take the chloroketone with TMSCl and Mg under Barbier-like conditions to the silyl enol ether and re-distill it, MeLi treatment at low temp could regenerate the enolate in a clean fashion.

    ReplyDelete
  13. I am covering this paper in my departmental seminar next week! Maybe I'll talk about some of the challenges listed here! Thank you for posting this! =)

    ReplyDelete