Multicomponent Synthesis of the SARS-CoV-2 Main Protease Inhibitor Nirmatrelvir

In the wake of the Covid-19 pandemic, it has become clear that global access to efficacious antiviral drugs will be critical to combat future outbreaks of SARS-CoV-2 or related viruses. The orally available SARS-CoV-2 main protease inhibitor nirmatrelvir has proven an effective treatment option for Covid-19, especially in compromised patients. We report a new synthesis of nirmatrelvir featuring a highly enantioselective biocatalytic desymmetrization (>99% ee) and a highly diastereoselective multicomponent reaction (>25:1 dr) as the key steps. Our route avoids the use of transition metals and peptide coupling reagents, resulting in an overall highly efficient and atom-economic process.


■ INTRODUCTION
The Covid-19 pandemic has impacted global health, society, and economies in a way unprecedented in modern history.Although large-scale vaccination against SARS-CoV-2 has brought the pandemic under control, serious concerns over the rise of new virus variants potentially evading an immune response remain.Consequently, there is an urgent need for orthogonal strategies to prevent or cure SARS-CoV-2 infection.One of the most promising therapeutic interventions to treat active infections with SARS-CoV-2, as well as related viruses that may emerge in the future, is the development of efficacious small molecule drugs.
Among the nonstructural proteins encoded in the SARS-CoV-2 viral genome, the 3C-like protease (3CL pro ) or main protease (M pro ) may be regarded as the most druggable target for the effective treatment of Covid-19. 1 Based on earlier work on the homologous SARS-CoV main protease, multiple highly active mechanism-based inhibitors were recently developed. 2he development of the orally available reversible covalent SARS-CoV-2 M pro inhibitor nirmatrelvir (1) by Pfizer was a major breakthrough in this area. 3In December 2021, the FDA approved Paxlovid (nirmatrelvir + ritonavir) for the emergency treatment of Covid-19 in the US.Regulatory approval in other countries (including the EU and the UK) soon followed.
Our groups share a common interest in developing efficient and sustainable methods for the production of complex small molecule drugs.Previously, we exploited our combined expertise in biocatalysis and multicomponent chemistry in the efficient synthesis of druglike prolyl peptides.Biocatalytic oxidation of meso-pyrrolidines with the monoamine oxidase N (MAO-N) D5 mutant afforded the corresponding bicyclic imines, 4 which were subjected to a diastereoselective Ugi-type three-component reaction (U-3CR) 5 to give bicyclic proline derivatives in high optical purity. 6This strategy was subsequently exploited in an efficient and diastereoselective synthesis of the hepatitis C virus (HCV) NS3 protease inhibitor, telaprevir. 7ith the emergence of new, highly active SARS-CoV-2 M pro inhibitors featuring bicyclic proline residues, 2d, 3 we recognized these compounds may be accessed using a similar strategy.In particular, we envisioned the rapid multicomponent assembly of nirmatrelvir from N-trifluoacetyl-tert-leucine (2), chiral bicyclic imine 3, and isocyanide 4 (Scheme 1).The latter could plausibly be derived from the known compound 5. 8 We anticipated that this U-3CR would proceed with high diastereoselectivity owing to highly effective shielding of the concave face of the bicyclic system of 3 by the gem-dimethyl moiety.We had previously demonstrated that the stereochemical outcome of such U-3CRs is determined solely by the chiral bicyclic imine, and not by chiral isocyanides or carboxylic acids. 6Biocatalytic access to imine 3 by MAO-Nmediated oxidation of 6 was previously reported by Merck and Codexis researchers, who also demonstrated it could be conveniently isolated as its crystalline bisulfite adduct 7 (Scheme 1). 9 This method was optimized for the large-scale production of 7 as an intermediate toward the HCV NS3 protease inhibitor, boceprevir (8), which features the same bicyclic proline residue as nirmatrelvir.Given the excellent accessibility of imine 3, we decided to pursue the multicomponent assembly of nirmatrelvir by U-3CR (Scheme 1) as a potentially greener alternative to the original route 3 (as well as other, related routes 10,11 ) that rely heavily on peptide coupling of the respective protected amino acids (see Supporting Information (SI) Scheme S1 and Table S1 for a direct comparison of our route and the original route).

■ RESULTS AND DISCUSSION
The bisulfite adduct 7 of the chiral imine 3 was prepared by monoamine oxidase (MAO-N) catalyzed oxidation of the amine 6 according to the published procedure. 9The enantioselectivity of the MAO-N-mediated formation of 3 (>99% ee) was determined by reaction of the cyclic imine 3 with phenylmagnesium bromide to yield the corresponding Grignard adduct, which was compared to a racemic standard (see the SI for further details).
Having secured access to key imine derivative 7, we next focused on the synthesis of requisite isocyanide 4. While it is widely accepted that the Ugi reaction and its variations typically have a very broad scope with respect to the isocyanide, there are only scarce reports of the use of chiral, highly functionalized isocyanides such as 4. 12 We have previously established that the generation of the isocyanide moiety from the corresponding formamide is compatible with the presence of esters and secondary amides. 7However, we soon ruled out the use of esters and nitriles as the X substituent in 4, as such α-acidic isocyanides typically display rather different reactivity. 13Moreover, α-isocyano esters tend to be configurationally unstable, while the near-complete absence of α-isocyano nitriles in the literature suggests they are inherently unstable. 14We thus decided to carry forward the C-terminus in a reduced form, i.e., as a protected primary alcohol.Given the demonstrated compatibility of esters with both isocyanide generation and the multicomponent reaction, 7 we selected 4a (X = CH 2 OBz) as the target isocyanide, with the benzoate facilitating TLC and SFC analysis.
The Boc-protected amino ester 5 has been reported by several sources 8 and is now also commercially available.Despite widely recognized reproducibility issues 8c with the key cyanomethylation step, 15 we were able to obtain sufficient amounts of 5 following literature procedures.8a,b We then turned our attention to the synthesis of isocyanide 4a (Scheme 2).Chemoselective reduction of the ester moiety followed by benzoylation smoothly afforded 10.Subsequent Boc cleavage followed by immediate formylation gave formamide 11 in excellent yield.The dehydration of 11 to 4a initially proved challenging, with most initial attempts leading either to incomplete conversion or to decomposition.Gratifyingly, under optimized conditions [triphosgene (0.66 equiv), Et 3 N (10 equiv), CH 2 Cl 2 , −78 °C] we could obtain 4a in 84% isolated yield.Alternatively, 11 can be dehydrated to give 4a in 83% isolated yield by treatment with trifluoroacetic anhydride (1.9 equiv) and Et 3 N (8 equiv) in THF (0 °C, 30 min.). 16To our delight, isocyanide 4a also proved to be a readily isolable, stable, crystalline solid.We were able to grow suitable crystals of 4a for X-ray diffraction and thus unambiguously confirmed its structure and absolute configuration (Scheme 2).
With all required fragments in hand, we moved on to the multicomponent assembly of the nirmatrelvir core.Initially, we attempted to use bisulfite adduct 7 directly in the U-3CR using model isocyanides and carboxylic acids as the other inputs.Although we were indeed able to isolate the corresponding Ugi adducts when the reaction was performed in protic solvents such as MeOH, the reaction was sluggish (plausibly as a result of the slow equilibrium between 7 and 3) and the isolated yield was low (∼20−35%).Moreover, we observed the formation of the "truncated" Ugi products (i.e., lacking the carboxylic acid moiety) as a side reaction. 17We therefore opted to perform the key U-3CR with free imine 3 (Scheme 3).With a moderate excess of carboxylic acid 2 and the volatile, in situ generated imine 3, the U-3CR smoothly proceeded to completion (i.e., with full consumption of isocyanide 4a) to give Ugi adduct 12 in 68% isolated yield.In our initial efforts, 12 was isolated as an ∼5:1 mixture of diastereomers (as determined by SFC-MS Scheme 1. Retrosynthesis of Nirmatrelvir and Biocatalytic Synthesis of Imine 3

Scheme 2. Synthesis of Isocyanide 4a
The Journal of Organic Chemistry analysis), which we initially attributed to incomplete stereoinduction at the newly formed stereocenter.Upon closer examination, however, we found that the procedure we used for the trifluoroacetylation of tert-leucine 18 led to some erosion of the stereochemistry, affording 2 in varying ee.Gratifyingly, the use of commercial 2 (>98% ee) in the key U-3CR afforded 12 in good yield with excellent diastereoselectivity (>25:1 dr).Subsequent methanolysis of the benzoate ester then gave alcohol 13 in a good yield.To our delight, oxidative conversion of the primary alcohol in 13 to the nitrile was smoothly accomplished using a convenient one-pot procedure 19 combining PhI(OAc) 2 /TEMPO oxidation with ammonium acetate as the nitrogen source to afford nirmatrelvir in 83% yield, with NMR data corresponding with literature data. 3ncouraged by the smooth conversion of essentially all of the steps in our synthesis, we investigated whether we could reduce the number of intermediate purifications (Scheme 4).
After streamlining some workup procedures, we were able to carry forward commercial ester 5 to isocyanide 4a without intermediate purifications with only minimal yield loss, affording 4a in 59% yield over four steps (versus 66% combined yield over the four individual steps).Advantageously, the subsequent U-3CR of 2, 3, and 4a can also be performed in MeOH, allowing the one-pot combination with the ensuing saponification to give 13 with >25:1 dr.The crude product 13 was then directly used in the final step, affording nirmatrelvir in 70% yield from 4a, which is considerably higher than the yield over the last three steps with intermediate purifications (48%, cf.Scheme 3).

■ CONCLUSION
We developed a new synthetic route to the SARS-CoV-2 M pro inhibitor nirmatrelvir featuring a biocatalytic desymmetrization and construction of the central bicyclic proline residue by an Ugi-type multicomponent reaction as the key steps.The process is scalable, efficient, and highly selective owing to the excellent enantioselectivity of the biotransformation (>99% ee) and excellent diastereoselectivity of the multicomponent reaction (>25:1 dr).Nearly all of the steps in the process proceed with full conversion, reducing the necessity for intermediate purifications.Despite a required reduction/ oxidation sequence, our route compares favorably to the original route in terms of step count and overall yield (SI Scheme S1 and Table S1).Combined, these credentials provide a basis for future development of the route to a possibly more sustainable and cost-efficient production process of this valuable API.
■ EXPERIMENTAL SECTION General Information.Compounds 2 and 5 were purchased from Angene Chemical.Other commercially available reagents were purchased from Merck (Sigma-Aldrich), Fischer Scientific, Strem Chemicals, TCI Europe, or Fluorochem and were used as received, unless mentioned otherwise.Solvents were purchased from VWR Chemicals or Sigma-Aldrich and used without purification unless stated otherwise.Anhydrous, air-free solvents (CH 2 Cl 2 , toluene, THF) were obtained from a PureSolv MD 5 solvent purification system.Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker Avance 600, Bruker Avance 500, or Bruker Avance 300 using the residual CHCl 3 signal ( 1 H: δ 7.26 ppm) or the CDCl 3 signal ( 13 C{ 1 H}: δ 77.16 ppm) as internal reference standard.Chemical shifts (δ) are given in ppm, and coupling constants (J) are quoted in hertz (Hz).Resonances are described as s (singlet), d (doublet), t (triplet), q (quartet), br (broad singlet), and m (multiplet) or combinations thereof.Structural assignments were made with additional information from gCOSY, gHSQC, and gHMBC experiments.Electrospray ionization (ESI) high-resolution mass spectrometry was carried out using a Bruker micrOTOF-Q instrument in positive-ion mode (capillary potential of 4500 V).Flash chromatography was performed on Silicycle Silica-P Flash Silica Gel (particle size 40−63 μm, pore diameter 60 Å) using the indicated eluent.Thin layer chromatography (TLC) was performed using TLC plates from Merck (SiO 2 , Kieselgel 60 F254 neutral, on aluminum with fluorescence indicator), and compounds were visualized by UV detection (254 nm), potassium permanganate, cerium(IV) sulfate, ninhydrin, and/or p-anisaldehyde stain.SFC-MS analysis was conducted using a Shimadzu Nexera SFC-MS equipped with a Nexera X2 SIL-30AC autosampler, Nexera UC LC-30AD SF CO 2 pump, Nexera X2 LC-30AD liquid chromatograph, Nexera UC SFC-30A back pressure regulator, prominence SPD-M20A diode array detector, prominence CTO-20AC column oven, and CBM-20A system controller, coupled to a Shimadzu LCMS-2020 mass spectrometer.The data were acquired in full-scan APCI mode (MS) from m/z 100 to 800 in positive ionization mode.Data was processed using Shimadzu Labsolutions 5.82.The Journal of Organic Chemistry polarimeter, using a 0.5 dm cell and solvent as indicated.GC-FID analysis was performed on an Agilent 6850 GC with a Gerstel Multipurpose sampler MPS2L using β-dex 325 (Supelco) with dimensions 30 m × 0.25 mm × 0.25 μm.X-ray crystallographic data were obtained from a Bruker D8 Quest instrument, equipped with PHOTON II detector.X-rays were generated from an INCOATEC IμS 3.0 Cu Kα sealed X-ray tube source at 1.54178 Å. Measurements were carried out at 100(2) K, using an Oxford Cryosystems CRYOSTREAM 800.
Experimental details for the biocatalytic production of imine 3 and its bisulfite adduct 7, additional optimization data for the dehydration of 11 to 4a, and experimental details for the streamlined synthesis of 1 are reported in the Supporting Information.