Thursday, April 21, 2016

Making C–C Bonds with the Simplicity of Making Amide Bonds

Despite all of the wonderful advances in organic synthesis, amide-bond formation is still the most widely used reaction among those making medicines.   This truth has led some to declare their embarrassment and depression. Medicinal chemist Derek Lowe has a nice post reviewing this state of affairs.  The authors of the J. Med. Chem. piece attribute the "over amide-ification" of medicinal chemistry to the following factors: "commercial availability of reagents, high chemoselectivity, and a pressure on delivery. "
What if one could piggyback on the popularity of amide bond formation – where people pay very little attention to things like the low atom-economy of coupling agents whose sole purpose is to expel water from an acid – and repurpose both the starting materials and reagents to instead accomplish a cross-coupling that would generate new C–C bonds with a loss of CO2?  Such a reaction would probably find some use if the conditions and setup were nearly as simple as amide-bond formation and the process were chemoselective.

This brings us to our latest work in the field of nickel chemistry which was published today in Science. As a complement to our Ni-catalyzed aryl–alkyl cross-coupling paper from February, this paper features a general and practical sp3-sp3 decarboxylative alkyl-alkyl cross-coupling. 

Thanks to 70+ years of amide bond formation, alkyl carboxylic acids are easily accessed from almost every commercial chemical supplier. They are everywhere. Considering their stability and general lack of toxicity, alkyl carboxylic acids should be ideal coupling partners in organic synthesis as the abundance of amides in Medicinal chemistry has taught us. In some ways they are really Nature's version of a boronic acid. Although the addition of a nucleophile or access to alternative oxidation states is known, utilizing simple alkyl carboxylic acids to perform decarboxylative cross-couplings is still rare. 

In a nutshell, we believe that the redox-active esters used in these couplings, incidentally the same types of activated esters used to make amides (HOAt, HOBt, NHPI, etc.), will provide chemists a new tool to re-evaluate the retrosynthetic analysis of complex natural products and bioactive molecules. Here’s a behind the scenes look at the invention of this method.

During our work on Ni-catalyzed aryl-alkyl coupling of aryl zinc reagents and alkyl carboxylic acids, we wondered if we could apply similar reaction conditions to forge sp3–sp3 bonds, historically a much more difficult problem than sp2–sp2 or sp2–sp3 cross coupling. The first bottle of diethylzinc I found was old and didn’t look too great, but the reaction worked even on the first attempt. Since we were interested in coupling more organometallic alkyl groups than just those that you can buy from a bottle (ZnEt2, ZnMe2, etc), we found that we could prepare dialkylzinc reagents from 2 equivalents of alkyl Grignard reagent or alkyl lithium reagent mixed with 1 equivalent of a ZnCl2 solution. This solution could directly be used for the cross-coupling reaction (we didn’t need to remove the Mg or Li salts formed from the transmetallation). The procedure is also simple to do (even one of our undergraduate students managed to prepare the dialkylzinc reagents and perform the decarboxylative coupling in his first trial).   

With this simple and practical protocol, I found that almost every possible dialkylzinc reagent (besides dibenzylzinc) worked great in the cross-coupling reaction. Moreover, this reaction was found to be general for primary and other secondary redox-active esters. After I showed the preliminary table to Phil, he looked at it for a few seconds and asked me: “As a molecule maker, don't you think that this methodology could change people’s thoughts about retrosynthetic analysis?” After thinking about this and running these reactions for the past few months, I am totally convinced that this methodology could provide the community a C–C bond forming reaction analogous in simplicity, utility, and practicality to classic amide bond formation. 

With Jacob and Pep’s great help, we finished more than 40 total substrates in two weeks. Some of the more interesting examples include the use of feedstock acids like adipic acid, tartaric acid, and 2-fluoropropionic acid.  We also sought out more complex coupling partners, like pharmaceuticals (such as atorvastatin, cetirizine) and complex natural products (cholic acid and biotin). Proximity to the Cravatt lab gave us lots of fatty acids to play with as well. Unlike alkyl halides, you can't throw a stone in a typical chemistry lab without hitting a carboxylic acid. Also a shout-out to Brad Maxwell of BMS for running the radio labeled example that demonstrates that one can exchange –CO2H for any radio labeled carbon source you like.

Some selected examples from ca. 60 reported.
Another defining feature of this work is the ability to couple α-oxy acids readily and efficiently. While most α-oxy-halogen substituted compounds are unstable, and thereby unusable in cross-coupling chemistry, α-oxy acids could smoothly convert to different types of ethers, some of which are hard to synthesize through traditional Williamson ether synthesis (which is incidentally another one of the most used reactions in Medicinal chemistry). Although α-oxy acids work well in this chemistry, one current limitation (that we are working hard to solve) is that it is difficult to couple benzylic acids as well as benzylic zinc reagents in this chemistry currently.

An alternative to the Williamson ether synthesis...

As the Baran lab peptide synthesis aficionado, Lara came on board to test the alkyl-alkyl cross-coupling reaction in one final chemoselectivity challenge: the direct modification of native peptides on the solid-phase.  After all, if one has a reaction that is supposedly as easy as amide-bond formation, it better work on solid-phase.  She prepared a handful of model peptides on Rink amide resin, each bearing a free side-chain (aspartic acid or glutamic acid) or a free α-carboxylic acid. The acids were activated on-resin as redox-active esters and then subjected directly to solid-phase nickel couplings with various diaklylzinc reagents. Before cleavage of the peptides from the solid support, all excess reagents were simply washed away, identical to the way excess peptide coupling reagents are washed away after amide bond formation on the solid-phase. In this way, we were able to repurpose age-old solid-phase synthesis techniques for late-stage, nickel-catalyzed C-C bond formation. Accessing the types of compounds shown below would otherwise require the synthesis of each unnatural amino acid (usually a multi-step synthesis!) and its incorporation into a new resin-bound peptide. We think the ease with which we can rapidly install valuable, non-native functional groups (e.g. alkenes for bioconjugation or peptide stapling) onto completely native peptides is one major advantage of the solid-phase method. Given the potential for late-stage peptide diversification, we also hope that this technique will find applications in both academia and industry for the rapid synthesis of peptide drug analogues. Finally it's interesting to contemplate the use of this method for iterative organic synthesis without having to invent any new equipment or building blocks...
If it's as easy as amide-bond formation it better work on a solid phase...
After we demonstrated the transformation on various primary and secondary carboxylic acids, including resin-bound peptides, we were curious to find if we could extend this work to tertiary carboxylic acids. Due to the difficulties associated with cross-coupling of tertiary electrophiles, we considered this a Holy Grail of cross-coupling chemistry. Fortunately, Pep and Shuhei found that 6,6’-dimethylpyridine was the sweet ligand for this transformation. With a few other changes to the reaction conditions, tertiary bridgehead acids could be used to obtain the desired products in decent yield; however, no product was observed for other tertiary radicals. Although we couldn’t use tertiary acids for a direct cross-coupling reaction with alkylzinc reagents, Chao came up with a brilliant idea to use the stability of the tertiary radical to our advantage. The hypothesis that we had is that in the presence of a radical acceptor, such as an alpha-beta unsaturated ester, we might be able to trap the tertiary radical and then subsequently do a coupling with an organozinc reagent, such as PhZnCl. Well, as it would turn out, this totally works in good yield! This three-component coupling works well for both normal, unstabilized tertiary radicals as well as α-heteroatom stabilized radicals.
A 3-component coupling that makes quaternary centers with great ease.

Similar to our previous Ni coupling reaction, this one is also easy to run; it doesn’t require rigorous drying of glassware, no fancy equipment is required, and you get to see dramatic color changes (from deep red to deep green)! Even for our peptide synthesis escapades, specialized equipment for solid-phase synthesis wasn’t required to perform any of the reactions. All peptide synthesis was done manually (though occasionally involving some creative use of the rotovap as a heating and stirring apparatus), and all you need to get started is a solid-phase reaction vessel and a suitable resin (e.g. Rink amide). 

On a less serious note, we were curious if we could take carboxylic acids that you find at the grocery store or hardware store, such as biotin (Vitamin H), cetirizine (Zyrtec), isobutyric acid (in Vanilla), and 2,4-Dichlorophenoxyacetic acid (2,4-D, an herbicide), and use these products directly in the cross-coupling reaction. 
Acids are literally everywhere. Some common products containing alkyl carboxylic acids.
As shown in this video, even an office jockey like Phil can get back in the hood and successfully make both an amide and a C-C bond. Click here to see which of the commercial acids in the picture above he chose to do the coupling on. The results are surprising...
Finally, I wanted to point out that the SI for this paper (ca. 460 pages) has been carefully crafted with pictures of everything, a FAQ section, extensive mechanistic discussion/experiments, and even a troubleshooting section. Phil's request to us was to formulate a document that would answer as many questions as possible and allow for extreme ease in reproducibility without any ambiguity and we tried our best to do that.  

Let us know if you have any questions and thanks for reading.


Wednesday, April 20, 2016

C–H Oxidation Meets Electrochemistry

When we look back at some of our lab’s terpene syntheses using a “two-phase” approach, it becomes abundantly clear that one oxidation reaction is used more than anything else: the allylic C–H oxidation. What isn’t obvious just by reading, though, is how unappealing these reactions actually are to perform. In the case of taxayunnanine D, our synthesis required 30 equivalents of our unique chromium (V) reagent, while a challenging oxidation en route to phorbol required 100 equivalents of CrO3! While these reagents served their purpose, we knew that the cost and toxicity associated with them would unquestionably preclude scaling up of these syntheses.
Terpene synthesis in the lab has relied on allylic oxidation for many many years.

This challenge presented us a unique opportunity to design a new system for allylic oxidation with the goal of a scalable and sustainable solution as reported today in Nature. Having explored some anodic oxidations a couple of years ago in both methodology and synthesis, the idea of allylic oxidation using electrochemistry was pretty appealing. As is almost always the case, someone had a similar idea during the Cold War – unfortunately, conditions developed back then were not general and led to poor conversions and yields in the reactions. We theorized that developing a variant of N-hydroxyphthalimide bearing electron-withdrawing groups would improve the reaction, and we found that using tetrachloro N-hydroxyphthalimide (Cl4NHPI) as a catalyst in the presence of a co-oxidant such at tBuOOH led to dramatically improved yields in these reactions.
Lots of optimization summarized here

One thing that we don’t really go into in the paper, is what it took to get these reactions to work from an engineering perspective. Initially we used graphite rod electrodes, these are pretty much big things of pencil lead that we purchase from Mcmaster-Carr (an online supplier of hardware stuff). At some point during the optimization process, we realized that while we were able to get beautiful conversion by crude NMR, the mass recovery for the reactions was often extremely low. After a ton of frustration, we reasoned that the lost mass was being absorbed into the graphite electrodes! We tried extensive washing of the electrodes, even resorting to soxhlet extraction for days, but these procedures were cumbersome and even with the extractions, we never were able to account for all of the material. After some more experimentation, we began to look at alternative electrode materials. While there are many examples of preparative organic electrochemistry using gold or platinum electrodes, these are quite expensive and we reasoned that even if Phil would let us buy them (he probably wouldn’t) their high cost would discourage any other labs or process groups from using our methodology. Glassy carbon, which is also often used as an electrode material, was also cost prohibitive. After searching through the chemical literature as well as scouring through random materials suppliers on the internet, we discovered RVC foam.
RVC electrodes, simple and cheap.
This stuff is pretty much an inexpensive source of glassy carbon, and is sold in sheets of various “pores per inch”. Since glassy carbon is impermeable to gasses and liquids, we figured that this should improve the yield of our oxidation and not suck up material into the electrodes. It took us quite some time from getting our first order of RVC foam delivered, to having our optimized electrochemical setup. The key thing we found was that for the reaction to proceed efficiently, we needed to maximize electrode surface area. Eventually we came to the setup we describe in the paper (and have laid out in detail in the Supporting Information). In summary, to put together the cell, we take a brick of RVC foam (100 ppi, costs ~$20 for a sheet from K.R. Reynolds co. which can make >10 electrodes) carve out two half-cylinder electrodes using a razor blade, and then solder connections to the top using stainless steel wire. The electrodes are placed in the cell (a test tube) and to keep the electrodes from touching we put a glass slide between them.

Admittedly, this setup is a little crude, and we are sure that a group with more engineering resources at their disposal could fabricate a prettier and perhaps more efficient device to carry out this reaction However, we are not an engineering group, and really wanted to develop a way to run this reaction using stuff that was inexpensive, and for the most part can be found in any synthetic organic chemistry lab. With optimized conditions in hand, we went on to explore the substrate scope. Methodologists like simple substrates, so we did those to make them happy. One thing we want to highlight is that the reaction tolerates tertiary allylic alcohols, and gives none of the allylic transposition product that is usually observed when oxidizing these substrates using traditional (chromium) methods.
Some selected simple compounds

What we were really interested in, however, were natural products. We bought, isolated, and synthesized as many terpenes as we could in as short a time as we could, hoping to oxidize them to enone containing terpenoids. We were able to oxidize several different terpene skeletons bearing a variety of olefin substitution patterns, electronic characteristics, and oxidation barcodes. We even found that the conditions were mild and selective enough for a fully deprotected glycosylated steroid (try that with chromium!). One of our favorite substrates was the oxidation of valencene to nootkatone (compound number 5) that had been previously reported using 15 equivalents of chromium! For those of you who have never been exposed to nootkatone, it is the primary fragrance component of grapefruit aroma, so anytime we would make it the smell of delicious grapefruit would permeate throughout the lab. We were also excited by this substrate since recently nootkatone has recently been show to be quite an effective mosquito repellant and a San Diego based company that produces nootkatone and valencene via fermentation had recently been purchased for $ 59 million!

Lots of natural products were prepared in this paper.
At some point during this project, we were approached by one of our collaborators, Asymchem (a CMO in Tianjin, China) who are interested in developing a platform to conduct reactions using electrochemical methods in order to decrease cost and environmental footprint. We gave them our procedures and some instruction of how they might carry out this transformation on a larger scale. In a few weeks time they were able to get this reaction to work 100 g scale! Perhaps equally cool was their crazy simple reaction setup, which consisted of two graphite plates submerged in an eight-liter beaker, contained in a bucket filled with ice! At the time we were trying to put the paper together Phil wanted to up the scale even more (can we do a mole?!?!), but the only limit (at the time) was the fact that the potentiostat asymchem had only went up to 1 A. In theory if a process group wanted to there is no limit to the scale of this reaction. 
This IS bucket chemistry...
It is easy to say that a reaction is “green” but we wanted to really show how much of an improvement our allylic oxidation was compared to the current methods that are used to run allylic oxidations on scale. Since we had been working with BMS process chem to use this reaction for some of their early-stage intermediates (Thanks Martin and Ke!), they provided some interesting metrics comparing our method for the oxidation of dehydroepiandrosterone acetate to two other literature examples that used chromium and ruthenium. The Process greenness score (PGS) is a rating used by BMS (other process companies use similar metrics) to evaluate the environmental impact of a reaction, taking into account the amount of waste the reaction generates (including workup) and the overall efficiency of the reaction. A higher % PGS indicates a “greener” reaction. The electrochemical reaction scores 56% vs. 32 and 37% for the chromium and ruthenium-based oxidations. As stated in the paper the PGS associated with our oxidation is in the range of standard amide bond forming reactions, as well as palladium-catalyzed coupling reactions. It is also worth noting that the PGS does not take into account the cost of disposal or the toxicity of the reagents, which means that compared to the chromium and ruthenium oxidations, our electrochemical reaction is even more of an improvement.
Metrics support the obvious conclusion: Electrons are better than chromium.
Overall this was a really exciting project to work on. Neither of us thought we would end up doing electrochemistry in the Baran lab, and although at the beginning it felt way outside of our comfort zone and chemical expertise, it is an interesting experience to do something completely different. One of the coolest parts of the whole thing was actually developing something that is scalable and sustainable that people actually might use. Phil had been alluding to this project during some of his recent ACS talks, and we have already been contacted by a number of process groups in the fragrance industry wanting to apply this chemistry to the manufacturing of a few chemicals. While electrochemistry might seem daunting, once you get over the initial learning curve there are a lot of really cool and practical applications, and hopefully you will see some more out of our lab in the future. We hope that other groups (academic or industrial) will begin to embrace this methodology, and electrochemistry in general, as an enabling tool for sustainably carrying out redox reactions in the lab, especially on scale. 

-Evan and Brandon