key: cord-0737997-r0jb4cbt
authors: Unoh, Yuto; Uehara, Shota; Nakahara, Kenji; Nobori, Haruaki; Yamatsu, Yukiko; Yamamoto, Shiho; Maruyama, Yuki; Taoda, Yoshiyuki; Kasamatsu, Koji; Suto, Takahiro; Kouki, Kensuke; Nakahashi, Atsufumi; Kawashima, Sho; Sanaki, Takao; Toba, Shinsuke; Uemura, Kentaro; Mizutare, Tohru; Ando, Shigeru; Sasaki, Michihito; Orba, Yasuko; Sawa, Hirofumi; Sato, Akihiko; Sato, Takafumi; Kato, Teruhisa; Tachibana, Yuki
title: Discovery of S-217622, a Non-Covalent Oral SARS-CoV-2 3CL Protease Inhibitor Clinical Candidate for Treating COVID-19
date: 2022-01-26
journal: bioRxiv
DOI: 10.1101/2022.01.26.477782
sha: ec809b02a9ee9b7d8b2f912d5d83f5f40fe53907
doc_id: 737997
cord_uid: r0jb4cbt

The coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has resulted in millions of deaths and threatens public health and safety. Despite the rapid global spread of COVID-19 vaccines, effective oral antiviral drugs are urgently needed. Here, we describe the discovery of S-217622, the first oral non-covalent, non-peptidic SARS-CoV-2 3CL protease inhibitor clinical candidate. S-217622 was discovered via virtual screening followed by biological screening of an in-house compound library, and optimization of the hit compound using a structure-based drug-design strategy. S-217622 exhibited antiviral activity in vitro against current outbreaking SARS-CoV-2 variants and showed favorable pharmacokinetic profiles in vivo for once-daily oral dosing. Furthermore, S-217622 dose-dependently inhibited intrapulmonary replication of SARS-CoV-2 in mice, indicating that this novel non-covalent inhibitor could be a potential oral agent for treating COVID-19.

The global coronavirus disease 2019 (COVID-19) pandemic, caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), continues to spread worldwide; more than 315 million people have been infected, and 5.5 million have died as of January 2022. 1 Because therapeutic options remain limited, oral COVID-19 therapeutics are urgently needed, especially for non-hospitalized patients, to prevent hospitalization and death. 2 5 S1 pocket and Leu-mimic hydrophobic moiety in the S2 pocket (Figure 2a) . Non-covalent small-molecule inhibitors, such as ML188, 12 ML300, 13, 14 and the 3-aminopyridine-like compound of the Postera COVID moonshot project, 15 ,16 exhibit similar pharmacophores, which form a hydrogen bond with the side-chain NH donor of His163 in the S1 pocket and have fitted lipophilic moieties in the S2 pocket (Figure 2b, c) . Additionally, the hydrogen bond with the Glu166 main-chain NH recognizes the P2 main-chain carbonyl of the substrate and is conserved in the known inhibitors. Given these interactions, we hypothesized that these three pharmacophores, i.e., the acceptor site with the side-chain NH donor of His163 in the S1 pocket, the lipophilic site in the S2 pocket, and the acceptor site with the Glu166 mainchain NH, play critical roles in small-molecule binding (Figure 2d ). 7 We performed docking-based virtual screening using the crystal structures of the 3CL pro and ML188-like non-covalent small molecules (Protein Data Bank [PDB] code: 6w63). 19 Compounds from the in-house library were docked, then the pharmacophore filter described above was applied to each docking pose, and the 300 top-scoring compounds were evaluated via enzymatic assays using mass spectrometry to avoid the false positives that frequently occur in fluorescence-based assays, giving some hit compounds with IC50 < 10 μM.

Optimization of the PK profile is a common challenge in drug discovery and usually takes time to overcome. Therefore, if possible, potency optimization of a hit compound with favorable PK profiles is likely the most straightforward way to meet the urgent need for an oral 3CL pro inhibitor. Further profiling of hit compounds revealed that one of the hit compounds, 1, could be a potential lead for this project because it displayed potent enzymatic inhibitory activity and favorable PK profiles with oral bioavailability (Figure 3 ). An enzymatic inhibition assay revealed that the IC50 value of 1 was 8.6 µM, and the in vitro metabolic stabilities of 1, measured after 30 min of incubation in human and rat microsomes, were 97% and 71%, respectively. An in vivo PK study in rats demonstrated that 1 had a favorable profile for the oral agent, oral bioavailability (F) of 111% and a low clearance of 7.3 mL/min/mg. 8 Figure 3 . Structure-based optimization of hit compound 1 and profiles of the compounds. a Cytopathic effect inhibition assay with Vero E6 cells expressing human transmembrane protease serine 2 (VeroE6/TMPRSS2). b % remaining in human liver microsomes (HLMs) after 30 min. c % remaining in rat liver microsomes (RLMs) after 30 min. d Total clearance, e Oral bioavailability. f Not tested. g Intravenously administered 0.5 µmol/mL/kg (n = 2), nonfasted; h Orally administered 1 µmol/5 mL/kg (n = 2), non-fasted; i Intravenously administered 0.1 mg/0.2 mL/kg (n = 2), non-fasted; j Orally administered 3 mg/2 mL/kg (n = 3), non-fasted. k Evaluated as S-217622 fumaric acid co-crystal form.

We resolved the X-ray complex structure of 1 with the protease (Figure 4a ). As expected, the binding mode of 1 in the X-ray structure was similar to that obtained in the docking 9 ( Figure 4b) . The S1 and S2 pockets were filled with the methyl-amide and 3,4,5trifluorobenzene moieties, respectively. The 4-difluoromethoxy-2-methylbenzene subunit was placed in the S1' pocket. The 2-carbonyl oxygen of the center triazine moiety formed a hydrogen bond with the main-chain NH of Glu166. The other side of the 4-carbonyl oxygen was bound in the oxyanion hole of the protease, which formed two hydrogen bonds with the main-chain NHs of Gly143 and Cys145. The methyl-amide moiety was placed in the S1 pocket, of which, the carbonyl oxygen interacted with the side-chain NH of His163. The protease exhibited an interesting conformational change in the S2 pocket; the side chain of the catalytic His41 was rotated and formed a face-to-face π interaction with the 3,4,5trifluorobenzene moiety of 1, whereas the docking pose predicted an edge-to-face π interaction (Figure 4c,d) . Along with the side-chain flip of His41, the 4-difluoromethoxy-2methylbenzene fragment was placed in a slightly different site compared with that of the docking pose, in which the ether oxygen of the P1' ligand formed a hydrogen bond with the main-chain NH of Thr26. An imine linker formed a water-mediated hydrogen bond with the His41 side chain, indicating its contribution to the affinity. was rotated to form a face-to-face π interaction with the 3,4,5-trifluorobenzene moiety of 1.

Docked structure is in lime green, and the X-ray structure is in cyan (1) and orange (protein residues).

Keeping the hydrogen bonds confirmed by the X-ray complex structure, straightforward multiparameter optimization was achieved starting from hit compound 1 ( Figure 3) . First, for a better fit with the S1' pocket, we optimized the P1' ligand while keeping the hydrogen bond with Thr26. As a result, compound 2, having 6-chloro-2-methyl-2H-indazole as a P1' ligand, displayed a 90-fold improvement in enzymatic inhibitory activity while maintaining the favorable DMPK profile. Next, the P1 methyl-amide moiety was replaced with a range of heterocyclic compounds, thus yielding compound 3, which eventually became the clinical 11 candidate, S-217622. S-217622 showed a biochemical activity of IC50 = 0.013 μM, an antiviral activity of EC50 = 0.37 μM, and preferable DMPK profiles for oral dosing, such as high metabolic stability (96% and 88% in human and rat liver microsomes, respectively), high oral absorption (97%) and low clearance (1.70 mL/min/mg) in rats ( Figure 3 , Tables S1-S3). Furthermore, S-217622 showed even better DMPK profiles in monkeys and dogs than in rats, with low clearance, long half-lives (t1/2) of approximately 10 and 30 hours in monkeys and dogs, respectively, and high oral bioavailability for all animals tested, suggesting its potential use for once-daily treatment of COVID-19 without requiring a PK booster such as ritonavir. Figure 5 shows the X-ray co-crystal structure of 3CL pro complexed with S-217622. In the S1 site, the 1-methyl-1H-1,2,4-triazole unit fit to the S1 pocket, forming a hydrogen bond with the side-chain NH of His163. The distinctive His41 flip observed in 1 was maintained in the S-217622 complex, and the 2,4,5-trifluorobenzylic moiety occupied the hydrophobic S2 pocket and stacked with the side chain of His41. The P1' ligand, 6-chloro-2-methyl-2Hindazole moiety held hydrogen bonding with the Thr26 main-chain NH and hydrophobic contact with Met49 as seen in the co-crystal structure of 1.

12 Figure 5 . X-ray co-structure of S-217622 (3) and 3CL pro . 3 is colored in orange and the protein is colored in gray. Water molecules are shown as red spheres. Hydrogen bonds are indicated as yellow dashed lines; π-π stacking is indicated as a cyan dashed line. Tables S2, S3 ). Because no significant mutations have been reported near the catalytic center of 3CL pro in these variants of concern, orthosteric 3CL pro inhibitors should be effective against all strains known to date. Antiviral activity of S-217622 against SARS-CoV (EC50: 0.21 μM, Figure 6b ) was also comparable to that against SARS-CoV-2, where the sequence homology of 3CL pro between SARS-CoV-2 and SARS-CoV was well conserved. S-217622 13 (3) also exhibited potent antiviral activity against MERS-CoV (EC50: 1.4 μM, Figure. 6c), HCoV-OC43 (EC90: 0.074 μM, Figure 6e ) and HCoV-229E (EC50: 5.5 μM, Figure 6d ). As described above, S-217622 displayed broad antiviral activities against a range of coronaviruses, suggesting possible applications of this compound or its derivatives for the next pandemic caused by future emerging coronaviruses. S-217622 showed no inhibitory activity against host-cell proteases, such as caspase-2, chymotrypsin, cathepsin B/D/G/L, and thrombin at up to 100 μM, suggesting its high selectivity for coronavirus proteases (Table 1) .

S-217622 (3) exhibited no safety concerns in vitro in studies involving ether-a-go-go-related gene inhibition, mutagenicity/clastogenicity, and phototoxicity (Table S4 ). 

The synthetic scheme for compound 1 is described in Scheme 1. Starting from the pyrazole derivative 4, cyclization with Ethyl isocyanatoacetate and CDI was conducted, giving 5 in 90% yield. Then, an alkylation with 5-bromomethyl-1,2,3-trifluorobenzene followed by introduction of a 4-difluoromethoxy-2-methylaniline unit, to give 7 (40% in 2 steps). The ester group in 7 was hydrolyzed and then amidated with methylamine, yielding 1 (58% in 2 steps). Compound 2 was synthesized similarly as shown in Scheme 2.

S-217622 (3) was synthesized as described in Scheme 3. Starting from known compound 

Here, we described the discovery of S-217622, the first non-peptidic, non-covalent, oral 3CL pro inhibitor clinical candidate for treating COVID-19. When we started this discovery program, most of the known inhibitors were peptide substrate mimetics with covalent warheads that bound covalently to Cys145 in the active site of 3CL pro . We assumed that these peptidic and reactive structural features would cause problems in the DMPK profile, such as low oral bioavailability due to low cell permeability, low metabolic stability, and low stability in the blood serum. Thus, we began the de novo search for non-peptidic 3CL pro inhibitors using the SBDD strategy to combat the current SARS-CoV-2 pandemic. Virtual screening followed by biological screening yielded several hit compounds with IC50 values <10 µM, 20 and one of these hit compounds, compound 1, showed a favorable DMPK profile for an oral agent. Using the X-ray co-structure, SBDD-based structural optimization enabled >600-fold activity improvement while maintaining a good DMPK profile; this ultimately yielded the drug candidate, S-217622 (3). S-217622 (3) exhibited a favorable preclinical profile as a once-daily oral therapeutic agent for COVID-19 with promising antiviral activities to known variants of concern, a long half-life in vivo, especially in monkeys and dogs, excellent oral bioavailability, and steep efficacy in an in vivo mouse model infected with SARS-CoV-2.

These favorable profiles prompted us to progress S-217622 to clinical trials, and studies are ongoing. 

To a stirred solution of 1H-pyrazole-1-carboximidamide hydrochloride 4 (53. 5 3-(tert-Butyl)-6-(ethylthio)-1,3,5-triazine-2,4(1H,3H)-dione (9)

Compound 9 was prepared according to the reported procedure. 21 A mixture of 10 (4.88 g, 13.08 mmol) in TFA (9.8 mL) was stirred at room temperature for 4 h then stood at the same temperature overnight. After concentration under reduced pressure, the residue was azeotroped with toluene and triturated with diisopropylether to afford 11

(4.01 g, 97%) as a white solid. 1 Rat PK studies. The animal study protocol was approved by the director of the institute after reviewing the protocol by the Institutional Animal Care and Use Committee in terms of the 3R (Replacement/Reduction/Refinement) principles.

Rat PK studies were done at Shionogi Pharmaceutical Research Center (Osaka, Japan).

Eight-week-old male Sprague-Dawley rats were purchased from Charles River Laboratories.

For oral administration, the dosing vehicle was dimethyl sulfoxide/0.5% methylcellulose (400 cP) = 1:4. The compound was orally administered at 1-2 µmol/5 mL/kg (n = 2) under non-fasted conditions. Blood samples (0. Virtual screening. As a target structure for virtual screening, we retrieved the crystal structure of the SARS-CoV-2 3CL pro in complex with a non-covalent inhibitor, X77 (PDB-ID: 6W63) 19 , from PDB. First, the structure was prepared using Protein Preparation Wizard 28 .

Missing atoms and side chains were added, and the ionization states of the amino acids were calculated using Epic 29, 30 . Hydrogen bond networks were optimized, and energy was minimized with a heavy atom restraint of 0.3 Å. All water molecules were removed from the crystal structure, and the docking grid was set to the center of the bound ligand of X77. An in-house compound library was preprocessed by Ligprep 31 before docking. Virtual screening was performed via Glide 32,33 in SP mode. The generated docking poses were filtered by the predefined pharmacophores using Phase 34, 35 . The pharmacophores were set as the acceptor sites with the sidechain NH donor of His163 in the S1 pocket, the lipophilic site in the S2 pocket and the acceptor site with the Glu166 main-chain NH. Finally, the 300 top-scoring compounds that matched all pharmacophores were selected for enzymatic assays. These 3CL pro were collected and mixed with thrombin His-tag at 4°C overnight to remove the Nor C-terminus. Thrombin-treated SARS-CoV-2 3CL pro was applied to HisTrap FF 5 mL (Cytiva) to remove proteins with uncleaved His-tags. The flow-through fraction was applied to a Superdex 200 16/60 (Cytiva) equilibrated with 20 mM HEPES (pH 7.5), 150 mM NaCl, and 1 mM DTT, and the fraction containing the major peak was collected.

Co-crystallization of SARS-CoV-2 3CL pro with compound 1 and 3 (S-217622), diffraction data collection, and structure determination. C-terminal His-tag free SARS-CoV-2 3CL pro protein (4.4 mg/mL) was incubated with 500 mM compound 1 for 1 h at room temperature, and the complexes were crystallized by sitting-drop vapor diffusion at 20°C.

The crystal of the compound 1 complex was grown with buffer containing 0.2 M ammonium citrate tribasic, pH 7.0, with 20% (w/v) PEG 3350.

N-terminal His-tag-free SARS-CoV-2 3CL pro protein (4.6 mg/mL) was incubated with 500 mM of S-217622 for 1 h at room temperature, and the complexes were crystallized by sittingdrop vapor diffusion at 20°C. The S-217622 complex crystal was grown with buffer containing 0.1 M BIS-Tris, pH 6.5, with 2.0 M ammonium sulfate.

X-ray diffraction data were collected using a Rigaku HyPix6000C detector mounted on a Rigaku FR-X rotating anode generator. Data were processed by CrysAlis Pro 36 . The structures were determined by molecular replacement using MOLREP 37 with the SARS-CoV-2 3CL pro -inhibitor complex (PDB-ID 6LU7) as a search model 38 . Iterative modelbuilding cycles were performed with COOT 37 and refined using REFMAC 39 . The data collection and structure refinement statistics were summarized in Table S5 . (Table S4 ) and crystallography data collection and refinement statistics (Table S5) .

The coordinates and structural factors of SARS-CoV-2 3CL pro in complex with 1 and 3 (S-42 217622) have been deposited into PDB with accession numbers 7VTH and 7VU6, respectively. Authors will release the atomic coordinates and experimental data upon article publication.

Y.U., S.U., and K.N. contributed equally to this paper. 

We are also grateful to all our colleagues who participated in the COVID-19 antiviral program at SHIONOGI: Yasushi Hasegawa

Masayoshi Ogawa for structural analysis of the compounds

Akira Ino, and Kenji Yamawaki for scientific discussion and advice; Takao Shishido, Keita Fukao

Junji Yamane for X-ray crystallography analysis

Chinami Nekomoto, and Kayoko Kanasaki for the safety studies. We also thank Shionogi Technoadvance Research Co., Ltd. for the compound supplies and technical support in the pharmacological studies

ac) for editing a draft of this manuscript

DMEM, dulbecco's modified eagle medium; DMSO, dimethyl sulfoxide

LCMS, liquid chromatography mass spectrometry; LHMDS, lithium bis(trimethylsilyl)amide

HRMS, high resolution mass spectrometry

COVID-19) Dashboard

WHO Target Product Profiles for COVID-19 Therapeutics for non-hospitalized patients

Characteristics of SARS-CoV-2 and COVID-19

The SARS-CoV-2 main protease as drug target

Proteases and protease inhibitors in infectious diseases

Postinfection treatment with a protease inhibitor increases survival of mice with a fatal SARS-CoV-2 infection

Potent SARS-CoV-2 Direct-acting antivirals provide an important complement to COVID-19 vaccines

Crystal structure of SARS-CoV-2 main protease provides a basis for design of improved α-ketoamide inhibitors

Potency, safety, and pharmacokinetic profiles of potential inhibitors targeting SARS-CoV-2 main protease

Discovery, synthesis, and structure-based optimization of a series of N-(tert-butyl)-2-(N-arylamido)-2-(pyridin-3-yl) acetamides (ML188) as potent noncovalent small molecule inhibitors of the severe acute respiratory syndrome coronavirus (SARS-CoV) 3CL protease

Discovery of N-(benzo[1,2,3]triazol-1-yl)-N-(benzyl)acetamido)phenyl) carboxamides as severe acute respiratory syndrome coronavirus (SARS-CoV) 3CLpro inhibitors: identification of ML300 and noncovalentnanomolar inhibitors with an induced-fit binding

Structure-based optimization of ML300-derived, noncovalent inhibitors targeting the severe acute respiratory syndrome coronavirus 3CL protease (SARS-CoV-2 3CL pro

Crystallographic and electrophilic fragment screening of the SARS-CoV-2 main protease

Open science discovery of oral non-covalent SARS-CoV-2 main protease inhibitor therapeutics

Design of wide-spectrum inhibitors targeting coronavirus main proteases

Structure of COVID-19 main protease bound to potent broad-spectrum non-covalent inhibitor X77

Analysis of SARS-CoV-2 variant mutations reveals neutralization escape mechanisms and the ability to use ACE2 receptors from additional species

Preparation of triazine derivatives and pharmaceutical compound that contains same and exhibits analgesic activity

Preparation of substituted pyrrolopyrimidinylacetamides as inhibitors of the BCL6 BTB domain protein-protein interaction and uses thereof

SARS-CoV-2 variants with mutations at the S1/S2 cleavage site are 50 generated in vitro during propagation in TMPRSS2-deficient cells

Development and evaluation of a novel loop-mediated isothermal amplification method for rapid detection of severe acute respiratory syndrome coronavirus

Dipeptidyl peptidase 4 is a functional receptor for the emerging human coronavirus-EMC

Rapid and automated tetrazolium-based colorimetric assay for the detection of anti-HIV compounds

Development of one-step, real-time, quantitative reverse transcriptase PCR assays for absolute quantitation of human coronaviruses OC43 and 229E

Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments

Towards the comprehensive, rapid, and accurate prediction of the favorable tautomeric states of drug-like molecules in aqueous solution

Epik: a software program for pKa prediction and protonation state generation for druglike molecules

Schrödinger Release 2019-4: LigPrep, Schrödinger, LLC

Glide: a new approach for rapid, accurate docking and scoring. 2. Enrichment factors in database screening

Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy

PHASE: a new engine for pharmacophore perception, 3D QSAR model development, and 3D database screening. 1. Methodology and preliminary results

PHASE: a novel approach to pharmacophore modeling and 3D database searching

What is this?" A structure analysis tool for rapid and automated solution of small molecule structures

MOLREP: an automated program for molecular replacement

Structure of M pro from SARS-CoV-2 and discovery of its inhibitors

Features and development of Coot

Refinement of macromolecular structures by the maximum-likelihood method

We thank all participants and investigator teams in the clinical study. We acknowledge the National Institute of Infectious Diseases (NIID), Dr. Kouichi Morita (Nagasaki University) and Dr. Bart Haagmans (Erasmus University Medical Center) for providing the SARS-CoV-