Savile, Janey, Mundorff, Moore, Tam, Jarvis, Colbeck, Krebber, Fleitz, Brands, Devine, Huisman, and Hughes. Biocatalytic Asymmetric Synthesis of Chiral Amines from Ketones Applied to Sitagliptin Manufacture. Science, 2010, 329, 305-309. DOI: 10.1126/science.1188934.
Here’s a short follow-up on this previous post, which covered a biocatalytic reaction developed by Codexis to make the key intermediate in the synthesis of Merck’s drug montelukast (aka Singulair). The 2010 Presidential Green Chemistry Awards were just announced, and the award for “Greener Reaction Conditions” went to Merck and Codexis for developing an enantioselective biocatalyst for the synthesis of sitagliptin, Merck’s blockbuster anti-diabetes drug (aka Januvia). This work is also the subject of a recently-published Science paper from Merck and Codexis.
The paper describes the development of an enzyme-catalyzed replacement for the final reaction in the synthesis of sitagliptin, in which a ketone functionality is converted into an amine. In the existing process route (which itself is extremely efficient), this conversion is accomplished by reductive amination: First the enamine is formed (dehydrositagliptin), followed by an asymmetric, rhodium-catalyzed hydrogenation. The only stoichiometric byproduct is acetic acid from the enamine formation, and the catalyst loading is very low (0.15% Rh, 0.155% ligand), so this step is pretty attractive from a green chemistry perspective. In fact, Merck won a Presidential Green Chemistry Award in 2006 for developing this reaction! Still, the reaction requires specialized high pressure hydrogenation equipment, and the addition of a solid adsorbent to remove all traces of the catalyst from the final product (amazingly they’re able to recover the rhodium from the adsorbent). And while the 95-97% ee is good, it still requires an extra recrystallization to upgrade to >99% ee.
This overall transformation can be catalyzed by the transaminase enzymes, which require the cofactor pyridoxal phosphate (PLP). Conceptually it’s pretty similar to the transfer hydrogenation from the last post. In that reaction, isopropanol acted as a reductant, and in this reaction, isopropyl amine acts as both the reductant and the amine source. The bulk of the paper details how they changed a transaminase with no activity towards the ketoamide substrate into one that could do the desired transformation in perfect yield and enantioselectivity at elevated temperature in the presence of organic solvents. This involved a great deal of active-site engineering to accommodate the non-natural substrate, as well as many rounds of directed evolution to improve yield, activity, enantioselectivity, and stability. In the end they got something that works really well: in addition to the excellent yield and perfect enantioselectivity, the reaction can be run at a pretty high concentration and at somewhat elevated temperatures. This last part is important because good volumetric productivity (grams of product per liter per hour) is necessary for these reactions to be economical on a large scale.
According to the paper the overall yield is ~10% higher, and there is a 19% reduction in waste compared to the rhodium-catalyzed process – I assume this is in large part because no recrystallization is necessary to upgrade the enantiomeric purity of the product. This should amount to a 26% reduction in the E-factor for this reaction. They also claim a reduction in manufacturing cost, and the process is certainly safer than doing a high-pressure hydrogenation on a large scale. Sitagliptin/Januvia is a billion dollar drug, and Merck is making lots of the stuff, so this chemistry could prevent the formation of a lot of waste.
As an afterthought to the sitagliptin chemistry, they also tried out some of the enzymes that they generated during their directed evolution studies on other ketones. They report a few reactions that proceed with excellent enantioselectivity, but they need to use a lot more enzyme and the yields aren’t as good. Not bad though, considering they were evolving the enzyme to work on a completely different substrate. I’m not sure how this method stacks up against existing methods for making these types of amines, but this might turn into a useful process sometime in the future.