Wednesday, April 23, 2014

Chlorination War Club




Ever wonder what kind of name is palau’amine?
Well yes… this alkaloid was isolated from a sea sponge native to the island nation of Palau.


But after a few minutes of googling I came across the word La’au palau or just palau and according to a Hawaiian dictionary is a type of weapon or “war club”.

And upon further inspection you may convince yourself of how the two guanidiniums resemblance jagged shark teeth or something close to that nature.

Working on making these polar natural products has always been a purification nightmare. Rotovaping water/acetonitrile hplc fractions on the daily definitely has to account for the majority of butt-kicking I (and many others) have endured from these nitrogen-war clubs, especially when trying to make more than just a few milligrams. I am still haunted by the days my co-workers gave me the stink eye for hogging rotovaps for 8+ hours and even resorting to unconventional air blow-down techniques (important) only to be made fun of L (ty dane).

Fortunately sticking to nasty polar molecules most likely means not many chemist have had the chance to explore their reactivity and stability and sometimes the butt-kicking pays off!

This story started like most other stories in Barnowl. PSB wanting to make natural products on [gram-scale]. In the process of optimizing the first generation route to key chlorospiroaminoketone 3, some key reactions were discovered. Of these, one was a chlorospirocyclization using this crazy cocktail of tBuOCl, TfNH2 then TFA, DMP. How did we get to this mix? Well TFA and DMP did the oxidation but what about TfNH2? So here is a lesson that I learned about “working dirty”. Unlike the post-doc I was working with, I purified every reaction as pure as I could so in case something didn’t work in the subsequent step I had one less variable to troubleshoot due to potential impurities… well this result changed all that. Guanidination of the corresponding primary amine using N,N′-di-Boc-N′′-triflylguanidine (Goodman’s reagent) furnished 1. Eager to push forward with very little material, the guanidylation reaction was essentially used crude for the next step… working with impure material has its perks.

To be honest we didn’t know what the heck TfNH2 was doing and we didn’t spend too much time thinking about it. All we knew was that it made the reaction cleaner looking and that it made us feel good. Nevertheless we stuck to the old “if it aint broke don’t fix it” mentality and got the JACS communication out the door. After letting all the smoke clear, we found what was really going on and realized that TfNH2 serve a good purpose after all. It’s proposed that tBuOCl reacts to give intermediate 6 and this is the actual chlorination species. After realizing this we decided to see if simple BisBocGuanidine could do the trick and found out that indeed it could but not as effectively…

It turns out TfNH2 was acting as a reductant for over-chlorinated species (such as chlorination of 2 or 3). In the model system [simple allylic guanidine: 1-(cyclopent-1-en-1-ylmethyl)guanidine)] this wasn’t a big deal but on the real system (1), the addition of TfNH2 was (as our gut feeling told us) essential (presumably due to alcohol or ketone reactivity toward over-chlorinated products).



So there you have it. Now that you know the background of this story lets fast forward some things… so we can chlorinate olefins. Cool!! Sorry PSB says boring. What we really need to focus is on heterocycles. After some side chain optimization we decided on reagent 7 (chlorobis(methoxycarbonyl)guanidine or CBMG aka “palau’chlor”). This guy was like NCS on steroids but without the nasty side effects that usually follow the use of such performance enhancing agents. So what do I mean by this?? If you look at substrates 10 and imidazo[1,2,-a]pyrazine (these two show representative contrast) you can see that most mild reagent conditions (practical) fail to give product for 10, while more aggressive reagents (reactive) decompose imidazo[1,2,-a]pyrazine. CBMG was the best of both worlds. You can read the paper to see the substrate scope as well as rate differences among other things.




The two points I will highlight in this blog post are what you guys will probably care about the most: 1.) purification 2.) solubility (solvent effects).

So who likes purifying? Not me (at least not anymore. Lol). Unless its necessary, really I shoot to get away with a work up alone … in cases where I can’t? Well I say the heck with the work up!! Just concentrate and load the reaction directly on silica gel (especially on large scale).

The solubility of the des-chloro 7 is poor in just about any solvent but worst in acetonitrile. All reactions typically start out clear (completely soluble in CHCl3) and end up very opaque/cloudy with fine white precipitate. Always you can filter this precipitate out and makes the work up or column easier but in cases where the reaction goes to full conversion (using 1.1 equiv reagent), this feature is awesome. In the [gram-scale] example below, the unstable chloroindole 22 was quickly purified by just simple filtration and isolated in quantitative yield.



Another interesting point is solvent effects. As you can see from the above table, solvent can have a dramatic effect on chlorination reactivity. For example, reactions using NCS/CHCl3 system had a dramatic effect when changing to NCS/CH3CN (entry 1B vs entry 2B) but overall it was found that this was not always the case as there were examples using NCS (compared to CBMG/CHCl3) that showed no improvement in yields (entries 9B-12B) regardless of solvent.


By the way it is of note that a substantial amount of effort went into compiling the final substrate table. We wanted to make sure that CBMG was indeed superior in reactivity to NCS (or other chlorinating agents) and this effect was not just a solubility effect. In some cases we selected conditions that worked best for NCS to compare against CBMG (using identical conditions). Therefore, a lot of screening took place. But at the end of the day this tedious work of “optimizing the enemy” was seen necessary to prove CBMG was having a real effect.

So palau’chlor ehy?? Lol.
Well this originated from PSB’s recent trip at the Dallas ACS. During his talk he used the name palau’chlor and soon after emails started coming in asking for palau’chlor… I had to admit it was catchy. Reminds me of selectfluor.

How about bromination? fluorination?? Yes we are still trying.
PS. That movie is awesome.



11 comments:

  1. If I remember correctly, there is a solvent compatibility problem of NCS with chloroform and DCM (and also with DMF and THF) because of NCS propensity to radical-chain chlorination of the solvent - but 1,2-dichloroethane, CCl4, AcOH and MeCN are compatible.

    Regarding the palau'amine model cyclization: Have you looked into an alternate mechanism?: Do you think it is possible that the N-chloro compound (6) is a kinetic sink product (that is a lousy nucleophile on nitrogen) and it actually resists the cyclization, but it can serve as a good chlorinating agent on the starting material (1)? If that was the case, then transforming all starting material (1) to N-chloro compound (6) shuts down the cyclization, unless you have an alternate Cl(+) acceptor like TfNH2 around - that can liberate few % of the starting material (1), and the system of (1) plus (6) eventually equilibrates the kinetic product (6) to cyclized thermodynamic product (2).

    I mention this possibility because I have seen a similar problem with ring chlorination of acetanilides and acetamidopyridines using tBuOCl in AcOH. The N-chloro-N-acetyl product was very stable, isolable and it resisted chlorination even under forcing conditions. However, a very slow addition of tBuOCl (addition funnel or syringe pump) - to make sure that N-chloro co-existed long enough with the starting acetanilide - has solved the problem.

    ReplyDelete
    Replies
    1. Hi milkshake,
      Not sure about the solvent compatibility (radical-chain chlorination) part. From the examples I found thru sci-finder searches (NCS/solvent) that were used to compare our regent to mostly used DCM as the solvent if not then it was usually MeCN or DMF. In the case with CBMG most substrates showed good reactivity in DCM or CHCl3 (comparable) but for substrates that failed to react in those chlorinated solvents using MeCN did the trick. From screening other solvents with CBMG it was found that there was a compatibility issue with protic solvents (MeOH) which seem to kill the reagent and when mixing CBMG in acetone there seem to be heat generation. So definitely solvent effects are an issue with NCS and holds true for CBMG as well.

      As for the alternative mechanism, definitely the chlorinated intermediate (6) is more deactivated toward chlorospirocyclization and although the addition of TfNH2 (or other reductant) can serve to free up some of the guanidine in your kinetic sink proposal, the addition of TfNH2 (using 1-2 equivalents of tBuOCl) with the simpliyed BisBoc version of allylic guanidine: 1-(cyclopent-1-en-1-ylmethyl)guanidine) was not necessary for effective chlorospirocyclization.

      In this simpler example the reaction was very fast and the only difference was dependent on the equivalents of tBuOCl used, as we can isolate a different ratio of overchlorinated products. Concentration studies were not done to try and see if we could say it was due to inter or intra chlorination but this model system showed that TfNH2 was not essential. In the real system (compound 1) I say TfNH2 turned out to be essential because without it there was more “tlc baseline”. Although it is hard to tell by the NMRs and have not isolated any byproduct to prove it, I have a feeling that the alcohol or ketone functionality is causing problems when TfNH2 is not around (since CBMG seemed to have some type of reaction with Acetone).

      The detail of these model studies as well as some other work regarding the axinellamines scale up will be published soon in a full article.

      Delete
    2. Rodrigo, thank you for your kind answer. Also, there is one more fairly practical chlorinating agent that is presumably stronger than NCS and tBuOCl; N,N-dichlorourethane aka DCU, EtOCONCl2, CAS# 13698-16-3; (Acros-Fisher sells it at good purity and price). I only used it for decarboxylative polychlorination of 5-methoxyindole-2-carboxylic acid, to the corresponding tri- or tetrachloro oxindole; I can't guess how well it would compare on other substrates, but you reagent has likely the advantage of ease of the byprodutct removal and a lower chance in participating in C=C addition of the N portion to olefin substrates. You have done an impressive reagent development work, I am sure CBMG will get used widely.

      PS: have you tried to combine your reagent with NaI in deuterated MeCN? Have you tried compatibility of CBMG with TFA (or even MeSO3H) - for chlorinating very deactivated substrates? Also, trifluoroethanol is hard to oxidize to an aldehyde, it might be worth trying, maybe it would be an exception from alcohol incompatibility.

      Delete
  2. Thank you milkshake. There are so many chlorinating reagents out there that the chlorination field seems like a hard area to make an impact but comparing CBMG to as many known reagents/conditions (both mild and aggressive) is the way we can really see CBMG's potential. We tried our best to compare reagents that people in lab have used (or liked the use of) so we will get our hands on some DCU and give it a go. Also we haven’t nailed down the reason for alcohol incompatibility so trifluoroethanol can serve as a good control.

    Considering the known procedure to make NIS with this system (NCS/NaI), yes it is on our list of things to investigate as well as the use of Br- sources. Preliminary studies toward the generation of other halogenated guanidines (at least using the same CBMG scaffold) has proved to be hard. So methods (even in situ generation) that will avoid the need to develop an entirely new guanidine scaffold are on the top of our priority list. Although the solubility of NaI is best in acetone the use of a different solvent is a worth trying.

    CBMG is compatible in TFA and I am sure it can react with deactivated systems. Unfortunately we observed little difference when comparing to the control NCS/TFA system. When using acids I feel the reactivity of the guanidine is lost due to different mechanistic considerations. In the NCS/acid literature I’ve seen proposals of Cl2 generateion, etc. Also I think that with TFA and other carboxylic acids you potentially can generate more reactive hypo-chloro type intermediates that can contribute to a loss in the “practical” aspect to the reagent so therefore we did not focus too much on the use of acids like TFA as additives/solvent.




    ReplyDelete
  3. 1)The supplementary info is not yet available in JACS available - Please, do you have X-ray structure of your reagent? It would be worth seeing how the carbamate ester groups are oriented in the crystal. Perhaps one of them can participates in the chlorination transition state as a hydrogen bond acceptor, helping to abstract H(+) from the substrate-Cl(+) adduct... The high reactivity of CBMG stands out - one would expect tBuOCl to be actually a lot stronger source of Cl(+) than NCS or CBMG simply based on electronegativity (and the fact that CBMG is made from tBuOCl), and also NCS ought to be a slightly more powerful Cl(+) source than CBMG because succinimide anion is a better leaving group than guanidinide dicarbamate ester anion... Something else must going on and your kinetic effect study suggests that facile H(+) abstraction is the key difference. And, the synchronous Cl(+) addition/H(+) abstraction would operate well in a weakly-coordinating solvent with high polarity, and your preferred solvent is chloroform, so this would be in agreement.

    2) A practical question about the HCl (1 eq.) additive - please have you looked into using TMS-Cl as the acidic additive, instead of HCl? Anhydrous HCl 4M solution in dioxane is unpleasantly corrosive and hygroscopic; TMSCl would be more convenient, i.e. when handled by Hamilton microsyringe.

    ReplyDelete
    Replies
    1. Dear Milkshake,
      Your milkshake brings all the boys to the yard (Damn right, it's better than Rodrigo's)

      Delete
    2. that thing, about bringing all the boys to the yard, did happen - last Thursday, when thermal runaway of my PEG polymerization sent a plume of boiling THF, ethylene oxide and polymer all over the hood... There was half kilo of ethylene oxide in the flask at the beginning, before the blow-up, and since EO is pretty toxic and flammable, everyone in the lab was dashing out in hurry. Also, scraping off and dissolving the solidified polymer goo that sprayed the manifold, stirplates and slimed up everything in my hood was what I have been doing the last two days - I just finished the clean up few minutes ago...

      Delete
  4. Its nice to see such interest and enthusiasm milkshake! For sure we are wondering this same point you bring up. We attempted to do some computational work to see if something like this could be possible but although we were able to correctly predict the conformation (compared to the x-ray data), we were unable to calculate a good transition state. The only thing we could find was a lower lying LUMO for CBMG vs NCS but this can also be explained by the small difference in N-Cl bond length. Ultimately the computational work failed to give us any useful data that could explain this reactivity differenece (CBMG vs. NCS and tBuOCl).

    Ill be glad to send you the cif files for both N-chloro guanidines we have crystal data on if you want to take a crack at it. Just shoot me an email at rodrigo@scripps.edu

    Also, I think tBuOCl is definitely reactive enough (since CBMG is made from tBuOCl as you point out) to chlorinate the substrates that CBMG can chlorinate (and tBuOCl failed). I wonder if maybe it's due to some activation barrier tBuOCl has to overcome and guanidine helps overcome this? Another point is mechanistic considerations as tBuOCl is believed to operate under a radical-based mechanism.

    As to the TMS-Cl, we did include that last part using HCl/dioxane as an additive but really this was just an extra feature that was not meant to be the main point of the reagent, especially since it taps into less-practical applications (as you point out).

    ReplyDelete
  5. Maybe my question is unexpected but I can not understand hiw did you come up with an idea of chloroguanidine as chlorinating agent?

    ReplyDelete
  6. Hey Zlatko,

    We came up with the idea from work we did on the axinellamines (http://pubs.acs.org/doi/abs/10.1021/ja206191g) here is the link to the synthesis paper if you do not have access you can shoot me an email and I can send you a PDF if you like. In that reference... the step in question is the one going from alcohol 20 to alpha aminoketone 24. We wanted to investigate the role of TfNH2 in the reaction.

    ReplyDelete