Based on the original research article published by Unoh et al. in the Journal of Medicinal Chemistry, together with the co-crystal structures 7VTH and 7VU6 deposited in the Protein Data Bank (PDB), this article describes the discovery process of S-217622. It focuses on how the researchers combined pharmacophore-based screening, co-crystal structure analysis, structural optimization, and drug metabolism and pharmacokinetics evaluation to optimize a micromolar-level hit compound into a clinical candidate with the potential for oral administration.[1–3]

1 Research Objective: To Obtain a Nonpeptidic, Noncovalent, Orally Available 3CL Protease Inhibitor

1.1 Role of 3CL Protease in Viral Replication

After severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) enters a host cell, its genome is first translated into polyprotein 1a (pp1a) and polyprotein 1ab (pp1ab).

The 3C-like protease (3CL^pro), also known as the main protease (M^pro), is responsible for cleaving pp1a and pp1ab at 11 sites, thereby contributing to the release of the nonstructural proteins required for viral replication. Inhibition of 3CL^pro reduces the production of mature nonstructural proteins, interferes with the formation of the replication–transcription complex, and consequently suppresses viral RNA replication.[1]

The catalytic center of 3CL^pro contains a catalytic dyad composed of His41 and Cys145. Its substrate-binding region includes subpockets such as S1, S2, and S1′. These subpockets recognize specific structural fragments of substrates or inhibitors and collectively determine the binding orientation and stability of the ligand.

1.2 Why Noncovalent Inhibitors Were Investigated

Many early 3CL^pro inhibitors employed peptidomimetic structures and contained electrophilic groups capable of reacting with Cys145. Such compounds can enhance target binding through covalent bond formation. However, peptide-like structures may be unfavorable for cell membrane permeability and oral pharmacokinetic properties, while the possibility of off-target covalent reactions associated with reactive functional groups also requires evaluation.

The S-217622 project was primarily developed around the following three objectives:

1. To employ a nonpeptidic small-molecule structure and reduce dependence on peptide-like scaffolds;

2. To inhibit 3CL^pro through hydrogen bonding, hydrophobic interactions, and aromatic interactions, without relying on covalent bond formation with Cys145;

3. To improve enzyme inhibitory activity while also taking cellular antiviral activity, metabolic stability, and oral pharmacokinetic properties into account.

Because noncovalent inhibitors cannot obtain additional binding stability from covalent bond formation, they must stably occupy the protein-binding pocket through multiple noncovalent interactions with appropriate orientations and distances.[1]

2 Hit Compound Discovery: Translating Protein-Binding Features into Screening Criteria

2.1 Three Key Pharmacophore Features

The researchers analyzed the binding modes of known 3CL^pro inhibitors and substrates and summarized the key recognition features within the protein active site into three pharmacophore requirements.

These requirements did not restrict candidate molecules to a specific chemical scaffold. Instead, they defined the principal spatial and interaction features that compounds were required to satisfy after entering the 3CL^pro binding pocket.

Using the complex structure of 3CL^pro with a noncovalent inhibitor as the basis, the research team performed molecular docking on hundreds of thousands of compounds from an internal compound library. A total of 300 compounds that satisfied the pharmacophore requirements and had relatively high docking scores were advanced to subsequent biochemical testing.[1]

2.2 Enzyme Activity Screening Using Mass Spectrometry

The primary screen employed a mass spectrometry-based enzyme activity assay rather than relying solely on a fluorescence reporter system.

Some compounds may interfere with optical assays through intrinsic fluorescence, fluorescence quenching, aggregation, or other mechanisms, thereby producing apparent inhibitory activity. Mass spectrometry directly detects changes in the substrate and enzymatic cleavage products, helping reduce false-positive results caused by optical interference.

The screening identified several hit compounds with half-maximal inhibitory concentration (IC₅₀) values below 10 μmol/L. Among them, compound 1 had an IC₅₀ of 8.6 μmol/L against 3CL^pro.[1]

3 Lead Compound Selection: Evaluating Both Activity and the Pharmacokinetic Foundation

3.1 Why Compound 1 Was Worth Further Optimization

Although the enzyme inhibitory activity of compound 1 was in the micromolar range, its drug metabolism and pharmacokinetics (DMPK) data provided a favorable foundation for further optimization.

Liver microsomal stability can be used to evaluate the susceptibility of a compound to transformation mediated by certain hepatic metabolic enzymes. A higher percentage remaining indicates that compound 1 was metabolized relatively slowly in the corresponding in vitro system.

Lower in vivo clearance is favorable for maintaining systemic exposure. An experimentally measured oral bioavailability exceeding 100% is an apparent pharmacokinetic estimate that may be affected by differences between intravenous and oral dosing experiments, sampling errors, clearance estimation, and other factors. It does not indicate that the actual absorbed amount exceeded the administered dose.

The principal value of compound 1 was that it already exhibited favorable metabolic stability, relatively low in vivo clearance, and oral exposure. The researchers therefore focused subsequent optimization on strengthening target binding while continuing to monitor whether the DMPK properties were adversely affected.[1]

3.2 Lead Compounds Require Comprehensive Evaluation

The selection of an oral candidate compound cannot be based solely on IC₅₀ ranking. The following factors must also be evaluated comprehensively:

① Whether the compound can enter virus-infected cells and reach the intracellular target;

② Whether oral administration can produce appropriate systemic exposure;

③ Whether the rates of in vivo metabolism and clearance are appropriate;

④ Whether the compound exhibits a certain degree of selectivity over other proteases;

⑤ Whether the chemical structure is suitable for further modification and the preparation of analogues.

Lead compound selection therefore represents a comprehensive assessment of target activity, metabolic stability, clearance, oral exposure, and structural modifiability.

4 Co-crystal Structure Analysis: Determining the Binding Basis and Optimizable Regions

4.1 Binding Mode of Compound 1

The researchers determined the X-ray co-crystal structure of compound 1 bound to 3CL^pro. The structure was deposited under PDB ID 7VTH at a resolution of 2.00 Å.[2]

Compound 1 contains a 1,3,5-triazine-2,4-dione core. The co-crystal structure showed that this core not only connects the different substituents but also directly participates in protein binding through its two carbonyl groups:

① One carbonyl group forms a hydrogen bond with the backbone NH of Glu166;

② The other carbonyl group is located in the oxyanion hole and forms hydrogen bonds with the backbone NH groups of Gly143 and Cys145;

③ The carbonyl group of the methylamide directed toward the S1 pocket forms a hydrogen bond with the side-chain NH of His163;

④ The 3,4,5-trifluorobenzyl moiety of compound 1 occupies the hydrophobic S2 pocket;

⑤ The ether oxygen directed toward the S1′ region forms a hydrogen bond with the backbone NH of Thr26.

Together, these interactions stabilize the binding conformation of compound 1 within the active site of 3CL^pro.[1,2]

4.2 Conformational Change of the His41 Side Chain Creates an Aromatic Interaction

The molecular docking model predicted that the aromatic ring of the 3,4,5-trifluorobenzyl moiety in compound 1 could form an aromatic interaction with His41. The co-crystal structure further showed that the side chain of His41 underwent a conformational adjustment, allowing its imidazole ring to form a face-to-face π–π interaction with this aromatic ring.

This result indicates that the 3CL^pro binding pocket is not completely rigid. After a ligand enters the active site, the side chains of certain amino acid residues can adjust their conformations, thereby forming interactions that are not fully described by a static docking model.

The co-crystal structure therefore provided two types of key information:

1. Confirmation of which hydrogen bonds and hydrophobic interactions genuinely existed and should be retained in subsequent design;

2. Identification of protein space that had not yet been fully utilized, providing directions for substituent optimization.

5 Structural Optimization: Improving the Fit within the S1 and S1′ Regions

5.1 Retaining the Central Scaffold and Key Binding Interactions

Subsequent optimization retained the 1,3,5-triazine-2,4-dione core, together with the major hydrogen bonds and S2 aromatic interaction confirmed by the co-crystal structure.

The researchers mainly optimized the peripheral substituents directed toward the S1 and S1′ regions. At the same time, they continued the design strategy of using a trifluorobenzyl moiety to occupy the S2 pocket and adjusted the positions of the fluorine substituents on the aromatic ring. Compound 1 contained a 3,4,5-trifluorobenzyl moiety, whereas the final compound, S-217622, contained a 2,4,5-trifluorobenzyl moiety.

5.2 Optimization of the S1′ Region: Introduction of a Chloro-Substituted Methylindazole

The substituent of compound 1 directed toward the S1′ region was a 4-(difluoromethoxy)-2-methylphenyl group. The researchers structurally replaced this region and introduced a 6-chloro-2-methyl-2H-indazole group into compound 2.

This structure can form the following interactions within the binding pocket:

① The indazole nitrogen forms a hydrogen bond with the backbone NH of Thr26;

② The 6-chloro-2-methyl-2H-indazole moiety forms a hydrophobic contact with Met49;

③ The overall shape provides an improved spatial fit within the S1′ region.

The enzyme inhibitory activity of compound 2 was approximately 90-fold higher than that of compound 1, while favorable DMPK properties were maintained.[1]

5.3 Optimization of the S1 Pocket: Replacing the Methylamide with a Methyltriazole

In the S1 region, the researchers replaced the methylamide of compound 2 with various nitrogen-containing heterocycles and ultimately selected 1-methyl-1H-1,2,4-triazole.

S-217622 was designated compound 3 in the original article. Its co-crystal structure with 3CL^pro was deposited under PDB ID 7VU6 at a resolution of 1.80 Å.[1,3]

The principal binding interactions of S-217622 can be summarized as follows:

S1 pocket

Triazole ring N in the (1-methyl-1H-1,2,4-triazol-3-yl)methyl moiety ── hydrogen bond ── side-chain NH of His163

S2 pocket

The 2,4,5-trifluorobenzyl moiety occupies the hydrophobic S2 pocket, and its trifluoro-substituted benzene ring forms a π–π interaction with the imidazole ring of His41.

S1′ region

The 6-chloro-2-methyl-2H-indazole moiety occupies the S1′ region. The N1 atom of the indazole ring acts as a hydrogen-bond acceptor and forms a hydrogen bond with the backbone NH of Thr26, while the chloro-substituted aromatic region of the moiety forms a hydrophobic contact with the side chain of Met49.

Central region

One carbonyl group of the 1,3,5-triazine-2,4-dione scaffold forms a hydrogen bond with the backbone NH of Glu166, while the other carbonyl group extends into the oxyanion hole and forms a hydrogen-bonding network with the relevant backbone amide NH groups.

The three peripheral structures are directed toward different binding regions, while the central triazinedione scaffold controls the spatial orientation of the substituents and participates in the hydrogen-bonding network with the protein backbone.

S-217622 occupies the active site of 3CL^pro through these noncovalent interactions and does not form a covalent bond with the catalytic residue Cys145.

6 From Enzyme Inhibitory Activity to an Oral Candidate Compound

6.1 Biochemical Activity and Cellular Antiviral Activity

S-217622 had an IC₅₀ of 0.013 μmol/L against 3CL^pro. Compared with compound 1, which had an IC₅₀ of 8.6 μmol/L, its enzyme inhibitory activity was improved by approximately 660-fold.

In a cytopathic effect assay using SARS-CoV-2-infected cells, the half-maximal effective concentration (EC₅₀) of S-217622 was 0.37 μmol/L.[1]

IC₅₀ and cellular EC₅₀ evaluate properties at different levels:

In a cellular system, a compound is also affected by membrane permeability, binding to proteins in the culture medium, intracellular nonspecific binding, metabolism, efflux, and other factors. Enzyme IC₅₀ and cellular EC₅₀ values should therefore not be regarded as directly equivalent.

6.2 Pharmacokinetic Properties

While improving enzyme inhibitory activity, S-217622 maintained high microsomal stability and further reduced in vivo clearance in rats.

The original article also reported that S-217622 exhibited high oral bioavailability in the tested animal species. The relatively long elimination half-lives observed in monkeys and dogs provided a pharmacokinetic basis for further investigation of a once-daily oral dosing regimen.[1]

These results indicate that structural optimization not only improved target inhibitory activity but also maintained metabolic stability and oral absorption while reducing in vivo clearance in rats.

7 Antiviral Spectrum and Protease Selectivity

7.1 Activity against Different SARS-CoV-2 Variants

S-217622 exhibited similar cellular antiviral activity against multiple SARS-CoV-2 variants tested in the original article, with EC₅₀ values ranging from 0.29 to 0.50 μmol/L. These variants included the Omicron variant tested at the time of the study.[1]

S-217622 targets 3CL^pro rather than the receptor-binding region of the spike protein. The variants tested in the original article did not cause a marked disruption of the principal active-site interactions required for S-217622 activity, and the compound therefore maintained similar activity in these experiments.

This conclusion is limited to the viral strains and experimental conditions evaluated in the original article and should not be directly extrapolated to subsequent variants that have not been experimentally assessed.

7.2 In Vitro Activity against Other Coronaviruses

The original article also evaluated the in vitro activity of S-217622 against other coronaviruses.

EC₉₀ represents the drug concentration required to produce 90% inhibition. Because the cell types, assay methods, and evaluation endpoints used in the experiments with different viruses were not completely identical, these values mainly demonstrate that S-217622 possesses a certain degree of broad-spectrum in vitro activity against coronaviruses. They should not be used for a simple direct ranking of relative potency.[1]

7.3 Selectivity against Other Proteases

At the highest tested concentration of 100 μmol/L, S-217622 did not achieve 50% inhibition of the various human proteases or the human immunodeficiency virus type 1 (HIV-1) protease examined in the study, including:

① Caspase-2;

② Chymotrypsin;

③ Cathepsins B, D, G, and L;

④ Thrombin;

⑤ HIV-1 protease.

These results support a certain degree of enzymatic selectivity of S-217622 for coronavirus 3CL proteases. However, they cannot replace broader off-target evaluation, safety pharmacology studies, or toxicological studies.[1]

8 In Vivo Efficacy in Mice: Plasma Exposure Correlates with Reduced Lung Viral Titers

The researchers evaluated the in vivo antiviral activity of S-217622 in a BALB/c mouse model infected with the SARS-CoV-2 Gamma variant.

The mice received one oral dose immediately after infection and a second oral dose 12 hours after infection at 2, 8, 16, or 32 mg/kg. Viral titers in lung tissue were measured 24 hours after infection.

The experimental results showed that:

① S-217622 reduced lung viral titers in a dose-dependent manner;

② The mean viral titers in the 16 and 32 mg/kg groups were close to the lower limit of quantification of the assay;

③ Plasma drug concentrations increased with the administered dose;

④ At doses of 16 mg/kg or higher, the simulated plasma concentrations remained above the protein-binding-adjusted cellular EC₅₀ for an extended period.

This experiment showed that increased plasma exposure to S-217622 and reduced lung viral titers followed consistent dose-dependent trends, providing pharmacodynamic evidence for its in vivo antiviral activity.[1]

It should be noted that the mouse efficacy study used two doses administered 12 hours apart. The potential for once-daily oral administration proposed in the original article was based mainly on the pharmacokinetic results obtained in monkeys and dogs, rather than being demonstrated solely by this mouse dosing regimen.

9 Discovery-Stage Synthetic Route: Supporting Analogue Preparation and Structure–Activity Relationship Studies

9.1 Synthetic Route of S-217622

In the original article, S-217622 was prepared in four steps from the known compound 3-tert-butyl-6-(ethylthio)-1,3,5-triazine-2,4-dione 9.[1]

Compound 9

│ 1-(Bromomethyl)-2,4,5-trifluorobenzene

│ Potassium carbonate, acetonitrile, 80 °C

│ Yield: 93%

▼

Compound 10: Introduction of the 2,4,5-trifluorobenzyl moiety

│ Trifluoroacetic acid, room temperature

│ Yield: 97%

▼

Compound 11: Removal of the tert-butyl group

│ 3-(Chloromethyl)-1-methyl-1H-1,2,4-triazole hydrochloride

│ Potassium carbonate, N,N-dimethylformamide, 60 °C

│ Yield: 45%

▼

Compound 12: Introduction of the (1-methyl-1H-1,2,4-triazol-3-yl)methyl moiety

│ 6-Chloro-2-methyl-2H-indazol-5-amine

│ Lithium bis(trimethylsilyl)amide, tetrahydrofuran

│ 0 °C to room temperature

│ Yield: 25%

▼

S-217622

Common abbreviations used in the route are as follows:

• MeCN: acetonitrile;

• TFA: trifluoroacetic acid;

• DMF: N,N-dimethylformamide;

• LHMDS: lithium hexamethyldisilazide, also known as lithium bis(trimethylsilyl)amide;

• THF: tetrahydrofuran.

S-217622 can also form a 1:1 fumarate salt with fumaric acid in ethyl acetate. The original article reported an isolated yield of 95% for this salt-formation step.[1]

9.2 Chemical Nature of the Final-Step Reaction

Compound 12 contains an ethylthio group on the triazine ring, which can serve as a leaving group during nucleophilic substitution. LHMDS can deprotonate 6-chloro-2-methyl-2H-indazol-5-amine or substantially increase its effective nucleophilicity. The resulting nitrogen nucleophile attacks the electron-deficient carbon of the triazine ring, followed by an addition–elimination process that replaces the ethylthio group.

This reaction is a nucleophilic substitution on an electron-deficient nitrogen-containing heterocycle. The ring nitrogen atoms and carbonyl groups in the triazinedione system help increase the electrophilicity of the corresponding ring carbon, enabling the amine nucleophile to complete the substitution reaction.

The original article mainly reported the reaction conditions, product structure, and isolated yield, and did not specifically investigate the detailed reaction kinetics or intermediates involved in this step.

9.3 Main Purpose of the Discovery Route

From the structure of the route, the three peripheral fragments directed toward the S1, S2, and S1′ regions are introduced stepwise, which facilitates the preparation of structural analogues and structure–activity relationship studies:

The original article reported a yield of 45% for the introduction of the methyltriazolylmethyl moiety and a yield of 25% for the final indazolylamine substitution reaction. Some intermediates and the final product were purified by silica gel column chromatography. This route was therefore used mainly for the preparation of candidate compounds and analogues during the drug-discovery stage.

A manufacturing process would require further evaluation of raw-material supply, impurity formation and removal, crystallization and isolation, operational safety, equipment compatibility, and batch-to-batch reproducibility.

10 Key Decisions in the Discovery of S-217622

The discovery process of S-217622 can be summarized in five consecutive stages.

This research process established a continuous chain of evidence comprising biochemical activity, co-crystal structures, cellular antiviral activity, DMPK properties, and in vivo efficacy in mice. It addressed questions at three levels:

1. Whether the compound could inhibit 3CL^pro through a clearly defined noncovalent mechanism;

2. Whether the compound could inhibit viral replication in infected cells;

3. Whether oral administration could produce drug exposure sufficient to support in vivo antiviral activity.

11 Chemicals Related to the Synthesis, Salt Form, and Analytical Studies of S-217622

Table 1. Target Compound, Salt Form, and Key Reaction Reagents

Table 2. Solvents for Reactions, Extraction, Salt Formation, and Separation

Table 3. Auxiliary Materials for Workup, Preparative Separation, and Analysis

Note: The products listed above are representative Aladdin products related to scientific research, synthesis, and analytical studies. Specific specifications, grades, suitability, and certificate of analysis information should be confirmed through the Aladdin website and the corresponding batch quality documentation. These products are intended for research use only and do not constitute recommendations for clinical medication, diagnosis, or human use.

References

[1] Unoh Y, Uehara S, Nakahara K, et al. Discovery of S-217622, a noncovalent oral SARS-CoV-2 3CL protease inhibitor clinical candidate for treating COVID-19. Journal of Medicinal Chemistry. 2022;65(9):6499–6512. doi:10.1021/acs.jmedchem.2c00117.

[2] Yamamoto S, Tachibana Y. The crystal structure of SARS-CoV-2 3CL protease in complex with compound 1. Protein Data Bank. PDB ID: 7VTH. Released April 6, 2022. doi:10.2210/pdb7VTH/pdb.

[3] Yamamoto S, Yamane J, Tachibana Y. The crystal structure of SARS-CoV-2 3CL protease in complex with compound 3. Protein Data Bank. PDB ID: 7VU6. Released April 6, 2022. doi:10.2210/pdb7VU6/pdb.

For more related articles, see below:

Three Main Differences Between AEBSF and PMSF Protease Inhibitors

Protease Inhibitors