Tuesday, May 16, 2017

Practical Oxidation of Strong C–H Bonds using Electrochemistry

We are excited that our latest paper “Scalable, Electrochemical Oxidation of Unactivated C–H Bonds” was published in JACS earlier today! This reaction enables the oxidation of a simple unactivated methylene to a ketone. Although one may think that this reaction is similar to methylene oxidation by TFDO, a great advantage is its scalability as exemplified by a 50 gram scale oxidation of sclareolide. In addition, due to the different oxidation mechanism, this protocol tolerates alkenes and azines, in contrast to TFDO. Details on reaction setup, optimization, and scope can be found in either the paper itself or the accompanying SI. Therefore, in this blog, I’m going to tell you the behind-the-scene stories for this reaction instead. All right—let’s C-Harge up and scroll down.
Our last paper in this field was the oxidation of allylic C–H bonds (E. J. Horn et al, Nature, 2016, 533, 77–81)—oxidizing unactivated C–H bonds, especially the methylene units, was therefore our next logical goal. But, of course, it wouldn’t be easy considering that there are only a handful of chemical methods for methylene functionalization. Since the TCNHPI mediator did not work for these strong bonds, Phil suggested that we look for a "super mediator", which overcomes the huge reactivity difference between allylic (BDE = ca. 83 kcal/mol) and methylene C-H bonds (BDE = ca. 95 kcal/mol). Is there a limit to what can be used as mediators? The answer was a big resounding NO—anything under the sun (as well as in the darkness) could be examined. To us, this is the beauty of electrochemistry, it allows one to oxidize/reduce numerous reagents to access high-energy radical species potentially of unique reactivity in a simple and sustainable way.
Not surprisingly, my first three months had gone by without any progress. I tried all species that could conceivably generate a reactive radical species, including various amides and alcohols (assuming starting materials may be coaxed into electrocution in a non-sober state). I started rummaging through our inventory to identify next candidates. As I was involved in the Palau’chlor project years ago as a visiting student, my guanidine samples from earlier years caught my eyes immediately. I tried all of them in our electrochemical setup: All mediators I examined in that series were dead. For a long time, I thought I was looking for a needle in the haystack but Phil was always encouraging. After a prolonged period of condition searching, I finally found that 3-aceclidine, as a mediator, promoted C-H oxidation of adamantane in the presence of HFIP with ammonium electrolyte. Later, both HFIP and the choice of electrolyte were found to be crucial.
Designing and crafting a home-made electrochemical apparatus, though laborious at times, kept my spirit up during the dark days. Below is my latest collection. Undoubtedly, the electrodes played an important role in the reaction development. We always avoided expensive elements (e.g., platinum) at the outset of the project—therefore, I doubt I will ever amass a complete collection of the periodic table; nevertheless, the future may take me to some less well-studied elements.  
With a “home-made” electrochemical setup, I naturally wanted to oxidize compounds accessible to the common household. For example, the precursor of one of our oxidation substrates was readily (and inexpensively) obtained from Whole Foods Market.
To properly put this method in context, we compared it extensively to the TFDO oxidation, hitherto the “go to method” for methylene oxidation in our lab. I prepared TFDO myself and tried to oxidize some of the substrates in the paper. This was one of my more dreaded times during the project. I have always heard terror stories from colleagues on TFDO—how everything must be pre-cooled to cryogenic temperature with extra precaution. I followed the prep on Openflask but soon found out that an open flask does not suffice for this process. Rather, one needs a sophisticated setup. In my own experience, the preparation of TFDO itself was not that cumbersome on a small scale, but the yield was very low (around 2%). In addition, this was totally unexpected—due to its low boiling point, taking this reagent with a syringe was a great challenge. Maintaining cryogenic temperature was absolutely critical as the TFDO will be evaporated or decomposed. I can’t imagine using TFDO on a really large scale. Meanwhile, our collaborators from Asymchem found that our electrochemical method was very amenable for scale up.

Perhaps we could do another race to justify the comparison. However, I have yet to convince a volunteer (myself included) to prepare TFDO again.

Finally, I would like to thank all people involved in this research, including our great collaborators Mike and Jeremy at Pfizer, and the very talented folks at Asymchem for the large scale reaction. Please let me know if you have any comments and questions!

Kawa
                                  

Wednesday, April 19, 2017

Decarboxylative Alkenylation

Our latest publication in decarboxylative cross-coupling of redox-active esters was published earlier today.  We will let you check out the publication and the SI (>450 comprehensive pages) for the details of our decarboxylative alkenylation, so this blog post will discuss the behind-the-scenes account of how the project developed.

This saga started over one year ago, in December 2015 at Pacifichem in Honolulu, Hawaii.  Inspiried by back to back talks from Prof. Larry Overman from UC-Irvine and TSRI’s very own Prof. Ryan Shenvi towards the synthesis of clerodane diterpene. Our group’s recent experience in cross-coupling and as well as our prior work on natural product synthesis led us to hypothesize that multiple clerodane natural products could be accessed from a common carboxylic acid intermediate in short order via a decarboxylative alkenylation. 

Initial scheme to access clerodane diterpenes

Shortly thereafter, we presented the above scheme to Phil, and he quickly realized that this strategy could be complementary to the tried-and-true sequence of ester reduction/oxidation/olefination (DIBAL/Swern/Wittig, for example).  Our previous decarboxylative cross-coupling methods were targeted towards medicinal chemists, but olefins are prevalent in natural products, and carboxylic acids (from esters) are common synthetic intermediates, so the marriage of these two entities could prove useful for synthetic chemists targeting natural products.

Phil wasn’t satisfied with just targeting a single class of natural products; he tasked us with determining if this disconnection proved strategic for a variety of natural product classes (other than the clerodane diterpenes).

We went back to our office, and within the day we found that many different classes of natural products could benefit from this strategy.  In particular, macrocyclic polyketides (cladospolides) could arise from difunctionalization of widely abundant tartaric acid.  For smaller natural products, this strategy could be employed as an alternative to OsO4 dihydroxylation chemistry, which has been used in the synthesis of these types of natural products.  At this point, Phil declared, “mission is go for launch.”
Natural products synthesized

We were shortly thereafter joined by Dr. Tian Qin and a fellow graduate student, Kyle McClymont. With Tian and Kyle’s help, we started initial investigations on the decarboxylative alkenylation and arrived at optimized conditions shortly thereafter. We found that this transformation works under both Ni and Fe catalysis and is compatible with organometallic reagents derived by a variety of conditions. Additionally, the activation and cross-coupling can be carried out in the same reaction flask.

Although our true interest was in synthesizing natural products, we conducted a substrate scope as a testing platform for the viability of this methodology. The scope is presented in the publication, so we don’t want to go into detail here.  However, we do want to graciously thank our collaborators at Asymchem in Tianjin, China, who tested and showed that our method was viable even on mole scale (ca. 620 grams) without any significant changes to the reaction conditions.  We also want to thank Ben and Scott from Bristol-Myers Squibb, as they found that our methodology could be used to access methyl ketones when traditional methods (such as Weinreb amides) fail.

(Left) Redox-active ester (Center) Mole-scale reaction (Right) Purified product (in a 500 mL flask)

A new graduate student, Kyle Knouse, joined us in the summer of 2016.  His previous work on lyngbic acid and related natural products allowed us to identify these targets as prime examples for this methodology. We were also joined by Dr. Lara Malins, our group’s peptide expert, and she showed yet again that these decarboxylative reactions can be conducted on peptide substrates. With the combined efforts of the team, we were able to access over 60 substrates and 16 natural products. If you are interested in trying out this reaction, please see the following flow chart (also found in the SI) to guide you in the best conditions for your particular coupling partners.

Flow chart user-guide for decarboxylative alkenylation

In addition to being field tested both at BMS and Asymchem, Phil himself wanted to see how this reaction compared to traditional carbonyl olefination methodologies. Undeterred by his recent humiliating defeat, Phil didn't lose his competitive edge and challenged the lab’s resident samurai warrior, Yuzuru Kanda, in a race to install an olefin on a glutamic acid derivative. Yuzuru was given the Wittig-based route and Phil was given the decarboxylative route.   Who won?  Watch the video to find out…

After completing the project, in true chemist fashion, we celebrated by going on a hike at Iron Mountain, which was followed by a beer at Nickel Beer Company in Julian, CA. 


Please let us know if you have any questions or comments regarding the work!  Thanks for reading!

Jacob and Rohan


Thursday, April 13, 2017

From One Acid to Another

Our recent work on decarboxylative borylation was published in Science today! Since Science’s new online research article format has allowed us to include many details in the paper as well as the supplementary information, I will avoid repeating what has been published online. Instead, I would like to share some behind-the-scene stories.
This project originated during a brainstorming session in Phil’s office in last May. Phil’s eyes were beaming with excitement, describing the prospect of converting alkyl carboxylic acids into alkyl boronates. I thought it was a workable idea but failed to comprehend the boss man’s enthusiasm—we could turn a “C” into a “B” but I would always want an “A”. Probably sensing my indifference, Phil told me to learn about Velcade where the boronic acid served as a pharmacophore. I was mesmerized by the mechanism of action; I was more thrilled by the fact that more than 15% of clinical drugs, such as chlorambucil, indomethacin and Liptor, contain the carboxylate moiety,. Could we possibly develop a method to convert all these bioactive carboxylic acids into their bioisosteric boron congeners? I was intrigued and embarked on the project.
Alkyl Carboxylic Acids are widespread in medicine
The initial condition screening took several painstaking months. To make a long tough story short (see our 400+page SI), I started the optimization from a secondary RAE, and it was found that the preparation of MeLi and B2pin2 ate-complex, and the addition of MgBr2 etherate were crucial. The pre-mixed MeLi and B2pin2 provided an activated ate-complex and allowed the coupling reaction to occur under mild conditions. The addition of MgBr2 etherate was thought to promote the transmetallation of boronate species onto the Ni center. However, when I applied the optimized reaction conditions to the primary RAEs, the major products were carboxylic acids (hydrolysis) and decarboxylative dimerization products. Interestingly, it was found that changing the ligand from di-tBubipy to di-MeObipy could increase the yield. We hypothesized (thanks to Pep for the insightful discussion) that the electron-rich Ni/ligand complex could facilitate the single electron reduction of NHPI as well as the combination of the primary radical with the Ni center. However, further increasing the electron density of the ligand didn’t help the reaction. At this point, I was glad that Jie joined me in the struggle and he found that the addition of DMF increased the yield further. Building on this finding, he came up with a set of conditions for tertiary RAEs.


Admittedly, Miyaura borylation (such as Fu’s elegant work) based on alkyl halides provides access to some of our targeted boronates as well—but we’re most interested in complex substrates such as amino acids whereby the corresponding halides are not viable coupling partners. It was found that the decarboxylative borylation also worked well on mono-Boc protected amino acids; however, since the electron-deficient nature of the boron atom enabled it to form a three-membered ring with its neighbor electron-rich nitrogen atom, such an adduct was prone to undergo C-B bond cleavage on silica gel and I couldn’t isolate the desired product. Lisa also found the same problem with α-oxy carboxylic acids. In an almost illogical move, instead of going back to simpler substrates, we started probing more challenging dipeptides. To our utter surprise, the borylation of the dipeptide worked very well; we were able to successfully synthesize two FDA approved drugs, velcade and ninlaro, both from simple native peptides!
After learning about our little success, Phil avidly asked me to do the borylation on vancomycin. After checking the calender to make sure that date was not April 1 and spending the next half an hour drawing out the structure of our desired borono-vancomycin, I approached Dr. Okano (Boger lab) who not only provided me with materials, but also offered lots of insights. To our delight, we were able to perform the borylation successfully on a vancomycin derivative. Although this transformation did not bring about an increase in antibiotic activity, we were excited by the reaction’s chemoselectivity.
Overall, primary (1º), secondary (2º), tertiary (3º), benzylic (including 1º, 2º, 3º), stabilized or non-stabilized, carboxylic acids on different ring systems, complex drugs (heteroatom-containing) and natural products could all be transformed to their boronic ester form smoothly using this reaction. Lisa and Jie also found that tertiary and secondary boronate esters also could be prepared from carboxylic acids directly when RAEs were generated in situ. The reaction conditions do differ depending on the nature of the substrates, but we provided a Guide for Selecting Reaction Conditions in our SI (page S31) to help the chemists who would be interested in this chemistry. The reaction setup is not complicated (done in three stages, including making the catalyst/ligand stock solution, preparation of the MeLi and B2pin2 ate-complex, and mixing them together with RAEs and MgBr2 etherate). The following video shows that Tony, our talented high school intern and a member of our team, can independently handle the decarboxylative borylation. To find out exactly who Tony is and see this simple reaction in action, see the video:

We were fairly content with the reaction scope but one question eluded us: can we use this reaction to exploit the untapped medicinal potential of boronic acids? At that moment, Phil had a conversation with “His Excellency” President Peter Schultz about our transformation. After a few days, Dr. Arnab Chatterjee in Calibr contacted us, enquring if we could prepare for them the following three boronic acids. The corresponding trifluoromethyl ketones are human neutrophil elastase inhibitors that have been tested in Phase II clinic trial for lung diseases; theoretically, the replacement of trifluoromethyl ketone with boronic acid could enhance the activity—however, preparations of these boronic acids are quite challenging by conventional means.
Translational chemistry with the Calibr team led by Arnab Chatterjee
Jie synthesized the first two molecules mCBK 319 and mCBK320 using our decarboxylative borylation in a few days, and the products were provided as single diastereomers. The initial hElastase inhibitory evaluation (IC50) showed that both mCBK 319 and mCBK 320 were more potent than their trifluoromethyl ketone forms, and the native peptide did not show any activity at all. With this result, we wrapped up the study and submitted our report to Science. It was only later when Calibr and Jie found the third molecule mCBK 323 was the most powerful (IC50 = 15 pM, Ki = 3.7 pM). The reviews were encouraging; nevertheless, they also suggested that the elastase studies may be published in a separate account as our current disclosure was too preliminary. To me, this wasn’t a bad news at all—after all, I could end up getting two publications from this project. However, Phil always values quality way more than quantity. The editor, Dr Jake Yeston, was also really accommodating and offered a second option that we expand the study into an online research article. Phil opted for this without any hesitation. Our colleagues from Calibr: Dr. Shan Yu, Dr. Kristen Johnson and Dr. Arnab Chatterjee worked tirelessly to complete a litany of additional biological assays in the narrow window of several weeks.
Scalable synthesis of a future clinical candidate?
We are also glad to report that mCBK 323 can now be accessed on gram scales; with abundant material supplies, in vivo biological studies are underway.
Our decarboxylative borylation is not without limitations. They are highlighted in the scheme below: 1) occasionally, the purification could be a problem, both due to the excess B2pin2 and the low boiling point of some simple pinacol boronate esters; 2) thioethers and thioesters were not well tolerated and led to low yields.

Limitations
Finally, I wanted to mention that in keeping with lab tradition, the Supporting Information is extensive. Phil told us that he wanted this document to be timeless so that if we all vanished in an earthquake or simply become too old to remember what we did, all the questions people may have and all the data needed to reproduce the reaction would be present for years to come.  

As always, if there are any questions please feel free to post them here or email us!

Chao and Borylation Team.