Synthesis of Kopsia lapidilecta alkaloids – The RCM approach takes a hit (retraction)

Let’s take a moment to appreciate the challenges of synthetic organic chemistry.  Not exactly stamping out widgets, is it?  Witness the recent retraction of an approach to the lundurine alkaloids from Steve Martin’s group (original Organic Letters paper here, retraction here).  Since the early days of RCM, Martin’s group has recognized its potential for the construction of alkaloids, but the presence of nitrogen atoms in RCM precursors can lead to problems.  Such may be the case here.

The Kopsia lapidilecta alkaloids

The Kopsia lapidilecta species of Asian flowering plants produces numerous alkaloids that bear the novel 5,6,12,13-tetrahydro-11a,13a-ethano-3H-pyrrolo[1′,2′:1,8]azocino[5,4-b]indole ring system.  Representative examples include lapidilectine B and lundurine B.

Synthetic efforts

Back at the University of Michigan, we became interested in making these alkaloids using our 2-azaallyl anion cycloaddition chemistry.  It was a war, but postdoc Ill Young Lee and Ph.D. student Patrick Stoy were up to the task.  The key step was the cycloaddition of the 2-azaallylanion shown with phenyl vinyl sulfide.  The resultant pyrrolidine cycloadduct was then converted on to lapidilectine B, completing the first total synthesis of any of the Kopsia lapidilecta alkaloids (JACS, JOC).

Martin’s group at Texas (link) and Sarpong’s group at Berkeley (link) have published approaches to the lundurines and lapidilectines, respectively, the former using RCM to make the central azocine core and the latter using an intramolecular electrophilic aromatic substitution approach.

Trouble with the RCM approach

Martin and co-workers reported a potentially simple way to access these alkaloids.  They carried out an RCM reaction on 24, which itself was made by an RCM assembly of the pyrroline ring.  The closure of the eight-membered ring was reported in 26% yield, producing 1.4 mg of 25.  A larger amount (12.9 mg) of 25 was then hydrogenated selectively to produce 26.  Their plan was to use this approach to make lundurine B. (Edit: Corrected yield of RCM.)


In a recent retraction, Martin has now withdrawn this work “on the basis that the RCM of 24 to give 25… is not reproducible; thus, the reduction of 25 to give 26… is also not reproducible.”  The wording of this sentence is curious to me, since the reproducibility of the reduction should not depend on the reproducibility of the RCM (the RCM product was used after purification), but nonetheless, this is a setback for this potentially powerful approach.

I hope the situation can be remedied.  Our synthesis of lapidilectine B was linear and involved a lot of steps, and I’d like to see the Martin or Sarpong approaches succeed; they have the potential to be considerably shorter.  I imagine there are other groups working on these alkaloids; they’re too beautiful to resist.  In my opinion.

Tags: , , , ,

11 responses to “Synthesis of Kopsia lapidilecta alkaloids – The RCM approach takes a hit (retraction)”

  1. Chemjobber says :

    It’s kind of a mysterious retraction, innit? I’d love to know what “not reproducible” means. It is entirely possible that it was made once, and never made again. Hmmm.

    • The Heterocyclist says :

      They must have run the reaction more than once. They made 1.4 mg in the RCM reaction (as reported in the SI), but hydrogenated 12.9 mg of this material in the next step.

      At first glance, I thought the problem might be the basic nitrogen and the adjacent quaternary center, but those problems also exist in the successful RCM that is used to make the pyrroline 24. Perhaps the additional constraints of the eight-membered ring formation tip the RCM into treacherous territory.

      • Chemjobber says :

        It doesn’t surprise me, certainly.

      • Anonymous says :

        Unless 25 was also made by another method (competing route?) and was then carried on to 26. It wouldn’t be the first time that a synthetic intermediate was made by a longer but easier (or cheaper) route in order to scale-up material.

  2. lennoxtutoring says :

    A 76% yield on a 1.4 mg scale is something I’d be very cautious about publishing. I imagine if the reaction were performed on a larger scale, the yield was probably lower, and hence not used as the reported yield in the paper. Otherwise, the reaction was limited to scales 26 was done on a scale of 12.9 mg indicates that 24->25 was done more than once. There is always the possibility that every aspect of the reaction was not studied sufficiently prior to publishing.

  3. Dave says :

    There’s a typo in your figure – Martin reported a 26% yield, not 76% yield. So, that’s 26% yield in that RCM (that lasted 5 days at RT) to give 1.4 mg from ~5mg of starting material.

  4. Ken says :

    This is why I love being a process chemist. You get to live with a reaction for a while and generally have enough material to figure things out.

    The step right before the “irreproducible” RCM is the exact same reaction conditions, just with less catalyst. They mention that when they attempted to do both reactions without stopping, the yield was lower. Says to me they likely have significant competing dimerization and/or polymerization. Figuring out the mass balance on 5 mg reactions in which dimerization and/or polymerization may be the main problem is pretty difficult, but that would be the first thing that would need to be done. Otherwise, you’re just going to be changing variables and hoping for the best.

    Here’s a couple shots in the dark in the absence of mass balance info:

    1) Consistent concentration

    Product versus polymerization would be dependent on that. My guess is they would test that variable.

    2) Consistent ethylene removal

    Reaction rates and extent of reaction could be significantly influenced by what is happening to the ethylene.

    3) Microwave, so consistent internal temperature for a consistent time

    The temperature and time design space of the reaction may be narrow, i.e. the window for having enough energy to get the desired reaction to occur while minimizing degradation could be small.

    4) Impurities

    Likely don’t have enough of the original lot that “worked” to tell if subsequent lots potentially had something in them that contribute to poor reaction performance.

    5) Less obvious variables

    Solvent variability, adventitious water, residual silica gel from purification, residual inorganics from purification, different batch of catalyst, variable solvent oxygen content, etc.

    So many variables, so little material and time.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: