Wednesday, October 26, 2011

Process Wednesday: That's a durn big molecule.

Wow. (Credit: C&EN)
This week's Chemical and Engineering News has a fascinating article on the bryostatins by Bethany Halford, complete with a little section on the process chemistry of Halaven:
Fang says that Halaven's potency and unique biological profile made it too attractive a target to pass up simply because it could be made only through a lengthy synthesis. "The perception wasn't so much that this was an obstacle but rather it was a challenge that we knew how to deal with," he says. 
And even though it takes a total of 62 steps to make Halaven, Fang points out that the synthesis is fairly convergent; the longest linear sequence is 30 steps. "The number of steps of a synthesis is one feature that people tend to focus on because it's easy to remember," he says. "But what really is critical to the successful implementation of a process in commercial manufacturing is not so much the number of steps but the types of purifications that are employed during the processing of the material." 
Chromatography, for example, takes a lot more time and generates a lot more waste than crystallization, Fang notes. "If you can take a 60-step synthesis and get rid of most of the chromatographies and replace them with crystallizations, then it's a much more manageable process than even a 10- or 15-step synthesis that has entirely chromatographic purifications," he says. 
Step count, Fang says, is nothing to be scared of. "Our feeling at Eisai is that natural products represent a large space of untapped potential new medicines," he says. "We're not deterred by a chemical obstacle. If the biological activity warrants it, we're more than happy to go after a compound."
Someday, an awesome set of Organic Process Research and Development papers is going to come out about this molecule. It will be great to hear the story of figuring out how to put this beast together. Until then, we'll just have to live on internet rumors and bang our heads on the wall that Eisai has only decided to publish in Japanese journals about the development of the process. 


  1. It's been my experience that Japanese chemists are very good at recrystallization. The comments by Fang seem to bear this out. I once asked a Japanese colleague her opinion on why this might be. She said it's probably out of necessity because the schools aren't particularly well-funded, and expensive silica gel is used as a last resort. I sure wouldn't mind taking a short-course on crystallization techniques from the Eisai process chemists. It's an art form, and would probably best be learned by apprenticeship.

  2. I thought they published the optimized process route in JACS few years back.

  3. All the SI is on milligram scale and published out of Harvard, not that they changed it all that much, I'll bet.

  4. As much as I love the Nozaki-Hiyama-Kishi reaction, the thought of removing all those chromium salts from a drug product gives me chills. According to an Eisai PowerPoint available online, the NHK methodology was key.

  5. jk: I don't mean any offense, but the "Crystallization is an art" meme needs to die. It's perpetuated out of ignorance. There are a number of books on the science of crystallization, and the Neal Andersen book CJ often pulls from for Process Wednesday has a chapter on crystallization theory.

    A crystallization can be developed systematically, just no one bothers to learn how.

  6. If I may, A2:53p, I offer a compromise between you and jk:

    Cheap, intuitive crystallization is an art.
    Expensive, instrument-analyzed crystallization is a science.

    Accept or reject the modification?

  7. Counter offer "crystallization can be done scientifically but often takes an artist to practice it well."

    I am not sure it as much no one bothers to learn as is so rarely taught in school. Column chromo has become a crutch to purify chemists don't get exposed to crystallization. My undergrad majors course included performing several crystallization's that were fairly easy (Aspirin is one I recall) but until I got into process did not fully appreciate the benefits and pleasures of developing a technique that yielded neat looking and highly purified material. And akin to formulations I was lucky to work with those who through extraordinary experience level were able to be excellent guides.

  8. (same as A2:53P)
    The crystallization studies that require turbidity probes to find the metastable zone and such are certainly out of the reach of an academic researcher, but a solvent cabinet will usually give you enough to work with to find a good binary system (forget ternary, that's too many variables). You can calculate solubility without any fancy equipment, just make a saturated solution, run it through a syringe filter, concentrate and weigh. Not that that's even necessary in most cases, but my point is there is a systematic approach whether or not you're going the whole nine yards.

    Maybe I'm just lucky (I know I'm not talented), but I haven't run into a solid that took so much effort to crystallize that I'd call myself an artist for getting it. Honestly, it just takes patience and sometimes many solvent screens to get some seeds, and from there you're usually golden. I think the patience part is what gets people.

    This also depends on whether you're after X-ray quality crystals or a well-behaved pure solid that won't clog a filter. Clearly, the latter is an easier goal, but single crystal quality can be obtained, again, by trying LOTS of solvent mixtures.

    Really, how is it different than getting a difficult reaction to work? You try lots of stuff and eventually it works. Or it doesn't. I don't see any difference.

  9. Anon 6:40, do you have a general set of guidelines you follow in determining crystallization conditions? Like if your compound is soluble in A, you can try to crystallize from X/Y/Z to start? Or if you have a favorite book you can recommend I'd appreciate that. I always find that I'm just guessing my way through a lot of these things and I could use a lot better education on it.

  10. Zoidberg:

    The basic antisolvents are either very polar (water generally) or very nonpolar (hexanes or slightly more polar depending on the solubility of the substrate). When possible, you want to to choose a solvent that boils lower than your antisolvent (ether/hexanes is usually where I start) because you can evaporate off the solvent faster than the antisolvent. If ether/hexanes doesn't work, it's often worthwhile examining alcohol solvents, acetone, THF etc. Benzene/cyclohexane is another combination that works pretty well for nonpolar molecules because you can take it to 0 without freezing the benzene. For more polar molecules, you often have to switch to something like acetone/ether, or as I said above, it's possible to use water as the antisolvent (but filtering aqueous solutions through a frit is a pain).

    Another thing that people maybe don't try are gentle stirring (makes xtallization more reproducible, but reduces size) and SLOW cooling. You don't want to go from boiling to the ice bath directly. Just leaving it in the oil bath to cool slowly is sometimes enough, but ideally you want to ramp it down slowly through the metastable zone.

    One thing to keep in mind: a huge part of process is keeping impurities out of the reaction mixture in the first place. That, and the fact that many of the intermediates are heterocycles, makes crystallization a good choice for purification method. However, getting pure material out of a complex mixture by crystallization is durn near impossible. You'd have to be very lucky (and working on a highly crystalline material) to be able to pull nice crystals out of sludge.

    As for books, I highly recommend Neal Anderson's Practical Process Research and Development. For, like, everything.

  11. Addition: seeding with a crystal grown from chromatographed material is another helpful idea. Knowing exactly when to add the seed can be tricky, but if you see nucleation, you've probably gone past the metastable zone. Warm it back to just above that and add some seed. If the seed dissolves, cool it just a little bit and try seeding again. Repeat until you see the seed remain undissolved and keep it at that temperature for a while, then cool slowly. Remember, too much nucleation = powder, so keeping it in the metastable zone to grow will generally give larger and better crystals.