By Michael Lambrecht
We all knew process scientists were good, but the Merck and Codexis teams just took it to another level! Unnatural nucleosides like Merck’s reverse transcriptase inhibitor islatravir (1, Figure 1) are well-known to be challenging to synthesize by traditional methods, as evidenced by several published chemical syntheses of this molecule requiring a dozen steps or more.1 In a recent Science article,2 Mark Huffman, Anna Fryszkowska, and colleagues conquered this challenge with a biocatalytic cascade in a single aqueous stream with no isolated intermediates, generating 1 in a whopping 51% overall yield as a single isomer which crystallizes out of solution. To accomplish this incredible feat, the team re-engineered a bacterial biosynthesis pathway to run in reverse, artificially evolving five different enzymes to do what they needed, and adding four auxiliary enzymes to drive the reactions in the direction they wanted. Wow!
The nucleotide salvage pathway in nature degrades natural nucleosides like cytidine into their component sugars and heterocycles. To re-engineer this pathway to synthesize the unnatural nucleoside islatravir, the Merck team needed to overcome two major challenges. First, they needed to edit the enzymes to accept their intermediates. Second, they needed to find a way to drive the equilibrium of the overall reaction in the opposite direction of the natural process.
To overcome the first challenge, they first generated a panel of phosphopentomutase (PPM) mutants to isomerize the phosphate from the 5’ position (compound 3, Figure 2) to the appropriate anomer at the 1’ position (compound 2). The mutants were generated based on their likelihood to impact activity based on a homology model. The mutants were tested for activity and mutations beneficial for activity were kept for a second round of evolution. The second round resulted in a PPM quintuple mutant3 with a 70-fold activity improvement over the wild-type enzyme for the unnatural ribose phosphate 3.
This PPM-based reaction could be coupled to a purine nucleoside phosphorylase (PNP)-based reaction employing an unnatural fluorinated base and unnatural phosphoribose 2 to give islatravir (1). The evolution process to obtain an improved PNP enzyme required more rounds of evolution to accept two unnatural substrates, but ultimately resulted in a PNP with 350x greater activity over the wild type enzyme. The two-step process created by evolution of the PPM and PNP enzymes also solved one of the biggest challenges in the chemical synthesis of the nucleoside: stereoselectivity. While the previous chemical synthesis4 struggled significantly with diastereomeric mixtures at the 1’ position, this evolved enzymatic process proceeds with >95.5:0.05 diastereoselectivity.5
With a process to form 1 from 3 in hand, a thermodynamic driving force was now needed to drive the equilibrium reactions toward 1, especially since the natural “salvage” process typically runs in the reverse reaction. This problem was solved by the addition of sucrose phosphorylase (SP) to remove the inorganic phosphate byproduct from the reaction mixture, allowing for high conversion to islatravir.
The re-engineering of the nucleoside “salvage” pathway is an impressive-enough accomplishment, but the Merck and Codexis team went further to come up with an enzymatic synthesis phosphate 3 from simple alkyne-triol 6 (Figure 3).
Here, the strategy was to evolve an oxidase to de-symmetrize the prochiral triol 6 to give aldehyde 5 and evolve a kinase to afford phosphate 4. To convert aldehyde phosphate 4 to ribose 3, a deoxy-ribose 5-phosophate aldolase (DERA) was used. Interestingly, while the wild-type enzyme catalyzed this reaction to some degree, high concentrations of acetaldehyde were not tolerated. So, the DERA enzyme was evolved to allow for acetaldehyde concentrations of >400 mM, a phenomenal feat considering aldehyde solutions are deliberately used to “fix” proteins and kill their activity!
Finally, some engineering was necessary to polish off the synthesis. Due to the polarity of some of the intermediates and the challenge of isolating and purifying the intermediates from aqueous solution, the authors wanted to carry a single aqueous stream through each step and crystallize the product from solution at the end. However, protein contamination of the final solid was problematic. To circumvent this, the oxidase and kinase were immobilized on solid phase to lower the amount of protein present at the end of the sequence and allow the final product to crystallize cleanly out of solution.
Altogether, five engineered enzymes were coupled with four auxiliary enzymes to give islatravir in 51% overall yield without isolating any intermediates. A summary of the overall process and the results of each enzyme’s evolution are summarized in Figure 4. This tour-de-force by Merck and Codexis truly shows the power of enzyme engineering and biocatalysis. Amazing.
Explore drughunter.com for more.
About the Guest Author – Mike Lambrecht
Mike Lambrecht is currently a Senior Scientist in the Discovery Chemistry department at Genentech. He was born and raised in central Wisconsin and received his B.S. in chemistry from the University of Wisconsin-Madison in 2010. In 2016, he obtained his Ph.D. from the University of Illinois at Urbana-Champaign under the direction of Prof. Paul Hergenrother. From 2016-2018, Mike was an Arnold O. Beckman Postdoctoral Fellow in the lab of Prof. Ryan Shenvi at The Scripps Research Institute in La Jolla, CA.
- I recommend reading the Merck process group’s excellent by-all-means chemical synthesis of this molecule to compare: McLaughlin, M, Kong, J. et al. “Enantioselective Synthesis of 4’-Ethynyl-2-fluoro-2’-deoxyadenosine (EFdA) via Enzymatic Desymmetrization.” Org. Lett. 2017, 19, 926-929. DOI: 10.1021/acs.orglett.7b00091
- Huffman, M. A., Fryszkowska, A. et al., “Design of an in vitro biocatalytic cascade for the manufacture of islatravir.” Science, 2019, 366, 1255-1259. DOI: 10.1126/science.aay8484
- i.e. with four amino acids mutated
- see Ref. 1
- This diastereoselectivity wasn’t a given at the outset, as the wild-type PPM enzyme produces product primarily of undesired stereochemistry.