Ester, Nitrile and Oxetane Play Nice

An useful paper published recently by the Pfizer team at Groton describes development of a robust process for the late-stage chemoselective saponification of an aryl ester present in the GLP-1R agonist danuglipron, a first-in class orally bioavailable molecule currently in clinical development for glycaemic control in patients with Type-2 diabetes (Figure 1).¹ The synthetic challenge here is overcoming competing hydrolysis of a remote benzonitrile substituent in the 2-fluoro-4-cyano benzyloxyl ring.² Danuglipron also contains an acid-labile oxetane ring. All of these “complications” made the key API-forming step, and in particular the control of impurities in the terminal deprotection, a focus for process development and optimisation. An outline of the earlier synthetic steps has been published.¹

Nitrile hydrolysis to amides and acids during cleavage of a carboxylic ester is a problem that is frequently encountered during synthesis. In the case of danuglipron, several approaches were screened, the most obvious of which being an enzymatic reaction. Insolubility of the substrate, however, rendered this approach impractical. Lewis acids and promotors such as I₂, LiI and LiCl and have been used successfully in the literature as promotors for chemoselective hydrolysis, however the oxetane ring in the substrate here was incompatible with these approaches.

The original method identified by the discovery team at Pfizer used 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD) in aqueous MEK (Figures 1, 2)- a reaction used in the literature to catalyse aminolysis of esters.³ TBD is a charge-neutral organic superbase with a pk_BH+ of 26. Its higher solubility relative to inorganic bases in lipophilic solvents makes it an attractive reagent and its often used catalytically in a wide range of industrial processes.⁴ A recent review by Fritz-Langhals describes the use of the cyclic base in more detail, including an improved and robust one-step method of synthesis.⁴

Initial efforts to scale the TBD reaction used in the penultimate synthetic step to  danuglipron focused on gaining more detailed process understating using a two-level full factorial design DOE (design of experiment) study. The parameters investigated were TBD (1-3 eqv), water (10-50 eqv) and MEK (5-15 ml/g) over a 20-60°C temperature range. Verification experiments were performed on large scale (100ml).

The experiments showed that the amide product concentration increased through hydrolysis of the nitrile as the ester hydrolysis reaction proceeded. Formation of the amide also correlated with high temperature and higher concentrations of water and base (Figure 3). Conversion plateaued at 90% under optimal conditions with 5% ester and 5% amide. The later could only be purged to acceptable levels by down-stream processing if the concentration was below 3%- a level obtainable if the reaction were stopped at 80% conversion. The limitations were that this would most likely prove difficult on scale and waste material in the API forming step.

Time for some innovative thinking. An alternative activation strategy comprising of conversion of the unreactive alkyl carboxylate to an activated ester followed by hydrolysis was inspired by a 2015 paper by Jamieson et alin which they describe increasing the reactivity of a methyl ester by transesterification with catalytic 2,2,2-trifluroethanol (TFE) and trapping with an amine to give an amide.⁵ By analogy, replacing the nucleophile with water would produce the acid. The Pfizer team quickly established proof of principle.

In the penultimate ester cleavage step towards danuglipron, use of K₃PO₄ as base in a TFE/water mixture gave a clean reaction profile (typically 93-94% acid, 2-3% of primary amide and 2-3% of starting material with <1.5% process impurities, Figure 1) superior original TBD process. The reaction kinetics proved sensitive to water content- with higher amounts leading to a faster reaction.

Another DOE full factorial design using volume of TFE, equivalents of K₃PO₄ and volume of water added as variables was carried out to investigate the reaction design space. Running under concentrated conditions was found to increase the rate of formation of the required product, however the amide product followed a similar kinetic pathway. From an operational perspective this would have had a significant effect on hold times at scale.

Using modelling studies, a region in which minimum water and base widens the reaction window by 10 hours was identified (Figure 4, 10g reaction).⁶ Isolated yields were around 82% and purity 97%. Recrystallisation from DMSO/water gave API within specification (88% recovery, 99.1% purity, 0.42% primary amide, 0.46% ester).

Reaction work-up and trisamine salt formation was carried out by adding water and toluene to the reaction mixture, with unreacted ester being rejected in the toluene phase. The product was retained in the basic, aqueous TFE phase. After pH adjustment with aqueous citric acid to pH 5 and extractive workup the solvent was exchanged to DMSO (TFE boils at 78°C). Trisamine was added and a temperature cycling crystallisation at 65-80°C API gave the trisamine salt.

The TFE exchange method gave an overall increase in isolated yield from 53% to 73%, and a process mass intensity reduction (PMI) from 455 to 60.

The use of fluorinated alkyl groups to increase the acidity of alcohols was reported as early as the 1930’s by Swarts. He described the synthesis of 2,2,2-trifluroethanol  via catalytic reduction of trifluoroacetic anhydride over a platinum-black catalyst.⁷ Reduction of trifluoroacetamide and trifluoroacetyl chloride proved more reliable methods. A review by Motiwala and Aube, ostensibly relating to the use of HFIP in organic synthesis, is a good source of information on the properties and synthetic applications of TFE.⁸ Both solvents have unique physical and chemical properties and are widely employed in organic synthesis.⁹ TFE is used to denature proteins to study folding and to stabilise the secondary structure of peptide solutions.¹⁰

My own experience with TFE was using it as starting material for the preparation of 2,2,2-trifluoroethyl difluoromethyl ether on route to Isoflurane anaesthetic. The reaction involved trapping difluorocarbene (generated from deprotonation of chlorodifluoromethane with NaOH and loss of chloride ion) with TFE (Figure 5). The alcohol is available on very large scale.

A good example of innovative process design from the Pfizer team here, solving a problem that many people will have faced or will face at some point during synthesis.

See you next time.

References

1. A small-molecule oral agonist of the human glucagon-like peptide‐1 receptor: Griffith et al, J. Med. Chem. 2022, 65, 8208−8226
2. Chemoselective saponification in the synthesis of danuglipron facilitated with trifluoroethanol: L. Han et al, Org. Process Res. Dev. 2024, https://doi.org/10.1021/acs.oprd.4c00085
3. A convenient aminolysis of esters catalyzed by 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD) under solvent-free conditions: C. Mioskowski et al, Tet. Lett. 2007, 48, 3863-3866
4. Unique superbase (TBD) (1,5,7-triazabicyclo[4.4.0]dec-5-ene): from catalytic activity and one-pot synthesis to broader application in industrial chemistry: E Fritz-Langhals, Org. Process Res. Dev. 2022, 26, 3015−3023
5. Catalytic amidation of unactivated ester derivatives mediated by trifluoroethanol: C. Jamieson et al, Chem. Commun., 2015, 51, 9495-9498
6. Kinetic and data-driven reaction analysis for pharmaceutical process development: J. Mustakis et al, Ind. Eng. Chem. Res. 2020, 59 , 2409−2421
7. The hydrogenation of trifluoroacetic and acetic anhydrides: F. Swarts, Soc. Chim. Belg. 1934, 43, 471−481
8. HFIP in organic synthesis: H. Motiwala et al, Chem. Rev. 2022, 122, 12544−12747
9. Fluorinated alcohols: a magic medium for the synthesis of heterocyclic compounds: S. Khaksar, Fluorine Chem. 2015, 172, 51-6; Fluorinated alcohols: a new medium for selective and clean reaction: D. Bonnet-Delpon et al, Synlett 2004, 1, 18-29; Fluorinated alcohols as solvents, cosolvents and additives in homogeneous catalysis: I. Shuklov et al, Synthesis 2007, 19, 2925-2943
10. Comparison of the effects of 2,2,2-trifluoroethanol on peptide and protein structure and function: M. Howard et al, J. Str. Biol. 2007, 157, 329-338